An exact fuel mixture in each cylinder is a critical condition of the engine performance. Only if the fuel mixture exactly corresponds to the required one, the engine will run evenly, its exhaust gasses will correspond the appropriate Euro norms, the fuel consumption will be minimal.
How does DME calculates, how much fuel should be injected in cylinders?
The modern DME use calculations of the energetic model of the engine. It means – several parameters are analyzed:
- inquired torque (the position of the accelerator pedal);
- air mass meter measurements (sucked-in air);
- intake manifold depression – indications of the pressure sensor;
- RPM of the engine and their changes, EGR, Tank ventilation system data, etc.
Using all these data, DME creates a virtual engine and, using complicated formulas, calculates a theoretically correct amount of fuel to be injected. Instead, the injector opening times can be calculated, if the pressure of the Rail and the flowrate of the injectors are known.
Unfortunately, each mechanical and electrical component has its technological tolerance – the difference of their parameters of the ideal. For some elements this technological tolerance is small, for example, +/-5%, for some – larger, and can reach several tens of % or even larger differences in specific conditions.
To compensate (avoid) these technological differences, DME creates multi-dimensional adaptation maps. These maps contain information regarding the difference of each parameter of the components from ideal parameters. During the time, these maps are continuously upgraded for the situation to correspond to the actual one, taking in account the aging of the elements and their changes due to this reason.
To control the result (the true fuel mixture), DME measures the content of the exhaust gasses in the exhaust system. For this reason, Lambda probes (oxygen sensors) are used. Of course, in a perfect case, the more efficient would be to use a separate Lambda probe for each cylinder. Unfortunately, this is a very expensive solution, so typically one Lambda probe for several cylinders is used. BMW uses the following strategy to control the exhaust gasses:
- N53 series engine: one wideband probe for every 3 cylinders;
- N43 series engines – one wideband probe for every 2 cylinders or all 4 cylinders (in “cheapest” N43 version).
Cylinders, which have a common Lambda probe, are called cylinders of one bank.
Main management principles
1. Wideband Lambda probes are performing the following functions (changing driving conditions; even mode, if no specific tests of the cylinder performance or other components are performed):
a) according to the signal of the probe, the difference between an ideal fuel mixture and the real situation is calculated. This difference is described as “Correction”. Correction – required correction for the fuel mixture (its average value in a range of the whole bank) to correspond the ideal;
b) from Correction data the “Integrator” is created. Integrator is a “slowed-down” Correction. This “slowing-down” is necessary to exclude very swift changes of the fuel mixture, which would create short-term unevenness of the engine performance. This Integrator inertia is parts of a second/seconds. Integrator data are immediately applied to the injector opening of all cylinders, included in the bank;
e) at the same time, the Integrators are used to create and correct the LTFT (Long Term Fuel Trim) adaptation maps. These changes are happening very slowly. Required time to change the adaptation maps – several tens of seconds. Also, these adaptation maps cover all cylinders of the bank – the fuel mixture is corrected equivalent in all cylinders.
Initial situation (correct; Lambda = 1.00; Correction/Integrator = 0%)
the calculated amount of the injected fuel by injector for cylinder No.1: 5.00 ml;
the calculated amount of the injected fuel by injector for cylinder No.2: 5.20 ml;
the calculated amount of the injected fuel by injector for cylinder No.3: 4.70 ml;
Suddenly DME has detected fuel leftover: 15%
Lambda, reported by the Lambda probe = 0.85;
Calculated Correction: -15%;
Integrator starts to move away from 0% (it becomes negative)
After 2 seconds the calculated Integrator: -10% (it comes close to -15%):
momentous changes, applied to the integrators:
5.00 ml – 10% = 4.50 ml;
5.20 ml – 10% = 4.68 ml;
4.70 ml – 10% = 4.23 ml;
After a second, Lambda, reported by the Lambda probe/confirmed Lambda = 0.95
DME has detected: fuel leftover: 5%;
After 3 seconds the Integrator reaches -15% and the fuel amounts, applied to the injctors:
5.00 ml – 15% = 4.25 ml;
5.20 ml – 15% = 4.42 ml;
4.70 ml – 15% = 3.995 ml;
Lambda, reported by the Lambda probe = 1.00 (ideal, corresponds the required):
DME has detected: fuel leftover: 0%;
Actual Integrator: -15%.
If this situation is stable – continues for a while (several tens of seconds), LTFT maps of the bank are modified, each injector is reducing the opening for 15%, after upgrade the Correction and Integrator aro getting close to 0 (if the required changes of 15% are included in the LTFT map).
More about STFT and LTFT read here.
2. Wideband Lambda probes are performing the following functions (even driving conditions: low/average/large load, if specific tests of the cylinder performance are performed):
a) wideband probe is measuring the changes in the fuel mixture when DME changes the opening time of injectors of one bank – gives information to DME regarding the chemical efficiency of each cylinder;
b) based on these data, DME measures each cylinder and in the exact driving conditions change the opening times of the injectors of measured cylinders in such way, so all cylinders would work with more or less similar efficiency;
c) as a result of these tests, the upgraded adaptation maps of each injector are created – their data are changed for the exact performance in the exact performance mode (torque, RPM).
The opening coefficient applied to the injector of cylinder No.1 in the exact conditions +3,2%
The opening coefficient applied to the injector of cylinder No.2 in the exact conditions +1,2%
The opening coefficient applied to the injector of cylinder No.1 in the exact conditions +4,7%
When performing the tests, DME detected the following differences of the injectors:
The opening coefficient applied to the injector of cylinder No.1 in the exact conditions +0,3%
The opening coefficient applied to the injector of cylinder No.2 in the exact conditions +0,6%
The opening coefficient applied to the injector of cylinder No.1 in the exact conditions -1,1%
As a result, the adaptation maps of the injector of each cylinder were upgraded in this mode:
the opening coefficient, applied to the injector of cylinder No.1: 3.2 + 0.3 = 3.5%
the opening coefficient, applied to the injector of cylinder No.2: 1.2 + 0.6 = 1.8%
the opening coefficient, applied to the injector of cylinder No.3: 4.7 – 1.2 = 3.5%
The new coefficients are recorded in the adaptation maps of the injectors and will be applied to future work.
3. Wideband Lambda probes are performing the following functions (even driving conditions: low/average/large load, if specific tests of the components are performed):
a) wideband probe measures changes of the fuel mixture in the exhaust;
b) DME maintains a correct fuel mixture in a range of the whole bank, using Correction/Integrator:
c) depending on the results of the tests of the components, upgrades their (not LTFT) adaptation maps.
Before the test of the fuel tank vent valve:
by TEV opening 5% expected fuel correction +3,2%
by TEV opening 10% expected fuel correction +4,3%
by TEV opening 15% expected fuel correction +6,7%
The required fuel mixture correction, detected in a result of the test:
by TEV opening 5% expected fuel correction +3,7%
by TEV opening 10% expected fuel correction +4,8%
by TEV opening 15% expected fuel correction +7,7%
New coefficients are recorded in the TEV adaptation map and will be used in future work.
4. Wideband Lambda probes are performing the following functions (idle):
a) wideband Lambda probe maintains a stable fuel mixture, while DME (using the data of the flywheel sensor) measures mechanical efficiency of each cylinder;
b) DME corrects the injector opening of each cylinder in such a way, so all cylinders would work more or less with the same efficiency;
c) if the differences of mechanical efficiency of cylinders are detected in the long term, the adaptation map op each cylinder is gradually modified to reach the most even idle possible.
Injected by injector of cylinder No.1: 5.04 ml;
Injected by injector of cylinder No.2: 5.37 ml;
Injected by injector of cylinder No.3: 4.27 ml.
Mechanical efficiency of cylinders (measuring the data of the crankshaft):
Cylinder No.1: +3,6%
Cylinder No.2: -1,4%
Cylinder No.3: -2,2% (the total average difference of cylinders 0,0%)
The 0.0% (perfect) efficiency of cylinders is reached if the injector injects:
The injector of cylinder No.1 injects 4.87 ml;
The injector of cylinder No.2 injects 5.44 ml;
The injector of cylinder No.3 injects 4.99 ml.
After a time (several tens of seconds/minutes, or when the idle mode session is finished) the new opening correction data are recorded in the adaptation maps of each injector and will be applied for the future work.
More about tests of mechanical efficiency read here.
5. Self-calibration/trimming of the wideband Lambda probe (even driving conditions, smoothed mechanical and chemical efficiency of cylinders, Homogeneous mode with a target Lambda 1.00):
a) DME changes fuel mixture in such a way, so the control probes would indicate the average voltage of 0.75V;
b) reaching the voltage mentioned, DME changes the adaptation map of the wideband Lambda probe in such a way, so its indication corresponds to Lambda 1.00.
More about management of Lambda probes read here.
Summary. As follows from the above:
a) in a “regular” mode and in the mode of changing obstacles (changing load, RPM), DME takes the planned amount of fuel of each bank from its adaptation map, regulates the fuel mixture of each bank, using Integrator data, changing the fuel mixture of all its cylinders proportionally; data regarding performance differences of each cylinder (its injectors) are taken from the adaptation map of each cylinder;
b) when the engine works in low/average/large load mode (by condition, that the fuel mixture of all bank is correct, the engine has no problem of system performance), DME can turn off the “regular” mode for some time and activate the tests of chemical efficiency of cylinders: during these tests, the injector adaptation maps will be upgraded to even the chemical efficiency of cylinders. When the chemical tests of cylinders will be completed (they usually last for several minutes), DME finishes these tests and switches back to the “regular” mode – controls the fuel mixture of the whole bank, using the probes;
c) when the engine works in the range of low/average/large load (by condition, that the fuel mixture of the whole bank is correct, the engine has no problem of the system performance), DME can turn off the “regular” mode for some time and activate the tests of another engine systems (EGR, TEV etc). In this mode, the wideband probe maintains the required fuel mixture in the range of the bank but modifies not fuel adaptation maps, but the adaptation maps of the system elements, which are tested;
d) in idle, if the DME has successfully stabilized the fuel mixture of the banks and the engine has no problems of another system performance, DME measures the mechanical efficiency of each cylinder (using the data of the flywheel sensor) and adjusts the opening of the injector of each cylinder in such way, so all cylinders would work more or less with a similar efficiency. The principle of theses tests is similar to the management of the fuel mixture of the banks – relatively fast the opening of each injector is changed: its “live” data, slower – the adaptation maps of each injector in idle are modified;
e) finally – the wideband probes itself are trimmed regularly, to ensure the most exact fuel regulation possible in each condition.