Cylinder Pressure Diagnostics in Large Engines. Influence Of

Home , Cylinder (engine), Cylinder (locomotive), Cylinder bank, Jenbacher

Cylinder Pressure Diagnostics in Large Engines. Get Better. With Kistler.

Influence of Crankshaft Torsion on Cylinder Pressure Diagnostics

Jürg Stadler, Mirko Ciecinski; Kistler Instrumente Herbert Schaumberger, Reinhard Steiner; GE Jenbacher Prof. Dr. Andreas Wimmer, Dr. Thomas Jauk; LEC/TU Graz

www.kistler.com Content

1 Motivation...... 3

2 Accurate Determination of TDC ...... 3

3 Integrating Torsion into Engine Process Calculation ...... 4

4 Simulating the Crank Mechanism ...... 4

5 Test Setup and Measuring Technology ...... 5

6 Measured Results ...... 6

7 Measured Results GETDC ...... 6

8 Validation of Crank Mechanism Simulation ...... 7

9 Combustion Analysis ...... 8

10 Influence of Crankshaft Torsion on IMEP ...... 9

11 Crankshaft Torsion and Energy Balance ...... 9

12 Summary and Conclusions ...... 9

13 Publication references ...... 10 Influence of Crankshaft Torsion on Cylinder Pressure Diagnostics.

Jürg Stadler, Mirko Ciecinski; Kistler Instrumente Herbert Schaumberger, Reinhard Steiner; GE Jenbacher Prof. Dr. Andreas Wimmer, Dr. Thomas Jauk; LEC/TU Graz

Torsion in long or flexible crankshafts can cause deviations A piston top dead centre (TDC) position that is too early, giving a between indicated and actual crankshaft angles at a given cylin- shift to the right on the cylinder pressure curve, leads to insuffi- der, and hence between calculated and actual cylinder volumes cient pressure when the piston ascends and excess pressure as it for a given crank angle. Kistler Instruments, GE’s Jenbacher descends. The result is apparent prolongation of the post com- product line and the LEC at Graz University of Technology have bustion phase and increased heat release. The circumstances are applied optical, in-cylinder measuring techniques to correct reversed when TDC is too late. The effects on angle offsets can cylinder volume during fired engine operation. The research has also be seen in terms of IMEP, which becomes greater when TDC led to more accurate combustion analysis, prediction of errors is too early and vice-versa. These effects are accompanied by and a method of verifying simulation results. This article is higher or lower friction mean effective pressures (FMEP). If a based on a paper presented at the 2014 Torsional Vibration maximum permissible deviation is defined as ±0,1 bar for IMEP Symposium in Salzburg (Austria). and ±1 % for the energy balance, then a precision requirement for TDC allocation of ±0,1 °CA can be derived. 1 Motivation In addition to high precision measurement data from cylinder 2 Accurate Determination of TDC pressure indication systems, ever greater requirements regarding Traditionally, TDC determination is based on calculations and thermodynamic analysis in engine development processes primar- measurements, either when the engine is at rest (static TDC ily demand precise knowledge of respective cylinder volumes determination) or non-fired engine operation (TDC determination during fired operation. via thermodynamic loss angle or by means of a capacitative sensor),as described for example in [1, 2]. The method of thermo- The prerequisite for precise calculation of cylinder volume is a dynamic adaptation is the only one that can also be applied in correspondingly exact measurement of crank angle. The signifi- fired engine operation. It is based on a calculation of the pressure cance of a crank angle deviation in thermodynamic analysis is curve during the phase without combustion and presumes precise depicted in Figure 1. The left diagram shows the basic influence knowledge, not only of the charge mass but also of heat transfer of a crank angle error on the energy balance of a large gas and leakage. The entire high pressure component can then be engine, while the image on the right shows the influence of an used in non-fired operation but, as a maximum, only in the range angle error on indicated mean effective pressure (IMEP). from intake valve closure to the onset of combustion in fired operation.

Figure 1: Influence of the crank angle error on energy balance and IMEP

www.kistler.com 3 In addition to the frequent non-availablity of gas exchange simu- sis, in order to be able to take into account the effects of load lation, which is a prerequisite for calculating the cylinder-specific related influences on the crank mechanism. The approach uses charge mass, this method leads to uncertainties, particularly in detailed simulation results of crankshaft torsion, validated with connection with the assumptions regarding heat transfer. More the aid of various measuring methods as additional input data in over, extreme valve timings, e.g. the Miller Cycle which involves the engine process calculation. The starting point is a crank very late intake valve closure, cause problems for TDC determi- mechanism simulation with which the load-dependent torsion nation in fired engine operation, since the intake range available curves on the individual crankpins were calculated as a function becomes very small. In principle, TDC determination in fired of crank angle. operation with all of the methods listed above is possible only under certain conditions, or not at all. The calculation results were validated with two different measurement methods in order to ensure their reliability. On the one The mechanically ideal engine takes into account neither manu- hand, measurements were carried out to determine the total tor- facturing tolerances nor deformations caused by thermal or sion of the crankshaft at its ends. On the other hand an optical mechanical load. In the case of real engines, and particularly with measurement system was used that was capable of determining large engines, considerable deviations from the ideal kinematics the gas exchange TDC on all cylinders durng fired operation. of the crank mechanism can arise in fired engine operation, due A validation of the calculated, cylinder-selective torsion curves at to the large dimensions and moving masses. The TDC values of a minimum of one point was then possible. This was concluded individual cylinders may thus deviate from the reference point by with an implementation of the results of real crankshaft a range of several degrees of crank angle. In addition, the piston behaviour into the engine process calculation. stroke no longer corresponds to the ideal, as crankshaft torsion changes throughout a full revolution. This results in a changed 4 Simulating the Crank Mechanism volume function. If such influences are to be taken into account The computing power available today enables an extensive simu- and integrated into cylinder pressure diagnostics, new methods lation of the crank mechanism [3]. The primary objectives of this must be applied that are capable of acquiring the real behaviour simulation are, among others: hardness analyses, determination of the crank mechanism under load, and of generating state- of resonance frequencies, design of the torsional damper and ments based on verified crankshaft simulation results. improving noise, vibration and harshness (NVH). Modern calculation tools offer the possibility of calculating the torsion of individ- 3 Integrating Torsion into Engine Process Calculation ual crankpins, as described in [4]. This information can be used, The method described below was developed and applied to a for example, for cylinder-dependent definition of the ignition tim- large bore gas engine for the purposes of thermodynamic analy- ing of an engine.

Figure 2: Angle of torsion of individual crankpins relative to a reference point

4 www.kistler.com Figure 3: Schematic measuring setup to determine the overall torsion

The crank mechanism is described using individual components In accordance with the method outlined above, the objective was such as crankshaft, connecting rod, piston, flywheel and torsional to measure the relative torsion on the flywheel in relation to the vibration damper. As the central structural element, the crank- crankshaft front end, in order to validate the crankshaft simula- shaft is broken down into a sufficient number of individual tion result. An optical crank angle encoder was used for this pur- masses, coupled by springs and dampers. The connecting rod is pose, in the range of the free shaft end (on the rigidly connected split into a rotating and an oscillating part. In the rotary oscillation part of the torsional damper). At the other end of the crankshaft model, the rotating connecting rod part is added to the corre- (flywheel side) the existing tooth ring was used with the corresponding crankshaft mass and the oscillating connecting rod part sponding pick-up sensor in order to determine the relative torsion is combined with the other masses moved by transmission (pis- of the crankshaft along its entire length, according to the method ton, piston pin, etc.), to form one oscillating mass (per cylinder). described in [5]. The signal of the encoder gave a resolution of The oscillating masses can be added to the rotating masses using 0,1 °CA, by means of which the torsion angles of the individual Frahm’s approximation. The loading and/or excitation of the sys- flanks of the tooth ring were determined, forming the reference tem are performed by cylinder pressure and the mass forces of for the measurement, as shown in Figure 3. As a result, crank the oscillating masses. angle-based data was available for relative torsion along the entire length of the crankshaft. Under the influence of gas forces and mass, vibrations are imposed on the crank mechanism. The movement of these vibra- In order also to obtain cylinder-specific measurement data, an tions is described via a system of differential equations. As a optical measurement system based on light conductor measure- result, the angle of torsion at a given computation point can be ment technology was used. This made it possible to acquire the represented in relation to a reference point. This function is used gas exchange TDC (GETDC) throughout the entire operating to determine the torsion of the individual crankpins in relation to range [6]. Using this measuring system, the measurement pro- a reference point, such as the crank angle encoder. The results gramme was carried out on each of the 24 cylinders in direct are shown in Figure 2. succession.

5 Test Setup and Measuring Technology The experiments shown in this article were carried out on a vee configuration, 24 cylinder large bore gas engine (unitary displace- ment above three litres) from GE Jenbacher. The measurement plan included operating points across the entire load range of the test engine at a nominal speed of 1 500 min–1.

www.kistler.com 5 Figure 4: Basic structure of the optical measurement system, incl. frontal illustration of the sensor

The system projects a laser beam into the combustion chamber of the piston, etc.) and on the other the flame propagation in the through an optical access point and detects the light reflected by "field of vision" of the sensor (flame type, intensity, etc.). On the the piston surface, as illustrated in Figure 4. The measured inten- engine measured, determining TDC under fired operation was sity of the reflected light is a dimension for piston position. This possible only in terms of GETDC. makes it possible to make qualitative statements about the piston stroke on the basis of crank angle and thus to determine TDC 6 Measured Results using the vertical section process [7]. A robust structure enables The engine was run at different loads at a constant speed of unlimited operation of the measuring system, even under fired 1 500 min–1. The curve depicted in Figure 5 for idle and full load engine operation. For the measuring campaign, the optical TDC resulted from the measurement of the torsion on the flywheel sensor was fitted practically flush mounted in the cylinder heads. relative to the crank angle encoder located at the front end ofthe Depending on the prerequisites that apply to the test engine, the crankshaft. As can be seen, the maximum torsion of more than GETDC can be determined with favourable accuracy (±0,2 °CA) 1,5 °CA appears along the entire length of the crankshaft at full in fired engine operation. Criteria for determining TDC under load. The amplitude of the torsional oscillation reduces as the fired engine operation are, on the one hand, structural conditions load declines. Observing the algebraic sign for relative torsion, it (sensor to piston distance at TDC, angle alignment on the surface is evident that, by and large, the flywheel side lags behind the front end.

Also depicted in this image is the simulation result of the total torsion of the crankshaft at full load, which exhibits excellent congruence with the measured results, not only in the absolute magnitude of torsion but also in the time curve over crank angle.

7 Measured Results GETDC To structure the depiction of the GETDC values measured with the optical TDC sensor more clearly, in Figure 6 separate presen- tations are displayed, according to the engine’s vee configuration, as Cylinder Bench 1 (cylinders 1 to 12) and Cylinder Bench 2 (cylinders 13 to 24). Once the crank angle encoder (reference basis) was installed on the front end in immediate proximity to the crankpin of cylinders twelve and 24, there was – as expected

Figure 5: Measured and calculated crankshaft torsion on the flywheel side, with – no TDC offset caused by torsion on these cylinders. However, respect to the front end at 1 500 min–1 the greater the distance from the front end of the engine, the

6 www.kistler.com Figure 6: Relative angle position of the GETDC for different loads with respect to the front end, optical TDC measurement system greater also the measured GETDC offset. Observations show that 8 Validation of Crank Mechanism Simulation the angles of torsion determined by the measurement technology A comparison of measured and calculated torsion across the applied increase progressively in the direction of the cylinders entire length of the crankshaft has been displayed in Figure 5. In nearest to the flywheel. This behaviour is clearly visible on both addition, the measured results from the optical TDC sensor cylinder banks, for which the absolute angle of torsion on Bench depicted in Figure 6 were applied for the validation of calculated 1 was generally somewhat greater than that on Bench 2. The crankshaft torsion. Figure 7 shows this comparison for full load reasons for this are structural and result from the mass relation- operation. Here, too, simulation was capable of reproducing the ships in the crank mechanism and the engine’s firing order. The measured data with a precision of better than than ±0,2 °CA on reduction in the TDC offset with decreasing load is also reflected all 24 cylinders. clearly in the measurement data.

Figure 7: Comparison of the angle of torsion in the GETDC with respect to the front end at 1 500 min–1 FULL LOAD; simulation vs. optical TDC measurement system

www.kistler.com 7 Figure 8: Resulting volume functions "baseline", "fully corrected volume function" and "cylinder-specific TDC correction"

9 Combustion Analysis • Method 2: Fully corrected volume function In order to be able to depict the effects of the inclusion of crank- The verified torsion curves from the crankshaft simulation were shaft torsion in cylinder pressure diagnostics, the LEC-CORA used for this method. In this case the torsion at the individual (zero-dimensional engine process simulation and analysis tool) cylinders, described above as ∆j(j), was taken into account in software developed at the Large Engines Competence Center the engine process analysis. (LEC) of the Graz University of Technology had first to be modi- fied and/or extended. The potential was created for overlaying • Method 3: Cylinder-specific TDC correction the crank angle dependent torsion curves with respect to individ- Because of the effort needed for method 2, a version was ual cylinders as ∆j(j) of the crank angle basis. From this there investigated which represents the measurement technology resulted a changed volume function, which formed the basis for determination of cylinder-specific TDC. For this purpose the the engine process analysis described below. corresponding values were obtained from the simulated torsion curves of the individual cylinders at TDC and used as TDC off- Three different methods were investigated in the context of the set values in the engine process analysis. subsequent comparison of results: The volume functions resulting from the three different proce- • Method 1: Baseline dures can be seen in Figure 8. The difference between the vari- Engine process analysis with the measured raw combustion ants "Fully corrected volume function" and "Cylinder-specific analysis data, on the basis of the cylinder-specific TDC deter- TDC correction" are depicted at ten times their actual size. mination over the thermodynamic loss angle performed on the test bench. As the full engine could not be run in non-fired The influences on IMEP and energy balance are depicted below operation, the non-fired cylinder pressure required for the in Figure 8 for the evaluation of the quality of cylinder pressure procedure was determined in a coast-down test. The range of diagnostics on the basis of the methods investigated. engine speed taken into account extended from nominal rated speed to 80 % of nominal rated speed.

8 www.kistler.com 10 Influence of Crankshaft Torsion on IMEP One initial interpretation of the comparison shown in Figure 9 is that the fluctuation range of IMEPs determined by the use of the raw measurement data (baseline) is comparatively large. A factor that has an effect on this fluctuation range is, certainly, the speci- fication of the charge mass for the individual cylinders. As only the entire mass can be measured on the test bench, the masses were distributed in equal parts over the individual cylinders. This assumption, which does not reflect reality, was followed for all the methods analysed.

Figure 10: Calculated energy balance, taking into account differently corrected volume functions at 1 500 min–1, full load

specific TDC correction" method result in a consistently lower percentage rate than with the "Fully corrected volume function" method, although the opposite picture is to be seen at cylinder Bench 2. Applied to the entire engine, the smallest range of fluc- tuations in the results occurred with the "Fully corrected volume function" method.

Figure 9: Calculated IMEP, taking into account differently corrected volume 12 Summary and Conclusions functions at 1 500 min–1 full load It was possible to demonstrate that crankshaft torsion is relevant to cylinder pressure diagnostics in large engines and that it can be If the volume function is corrected in accordance with Method 2 incorporated into their analysis. Of prime importance is the (fully corrected volume function), it will result in a distinctly more dependency of crankshaft torsion on givne operating points in ho-mogeneous distribution of the cylinder-specific IMEP values. terms of load and speed. This applies also for Method 3, when the calculated torsion error at TDC (cylinder-specific TDC correction) is used. This suggests The basis for the inclusion of crankshaft torsion into cylinder that good results can be achieved with this method, at least with pressure diagnostics is the crank angle-dependent torsion of the this test object. Depicted on the far right of the diagram is the individual crankpins, as determined in the simulation of the crank cylinder mean for the three methods. Here, deviations arise in a mechanism. A validation of the values for torsion can be achieved non-negligible magnitude of 1 to 2 % between the Baseline by the measurement of the total torsion of the crankshaft method and the virtually identical results from the two other between the flywheel and its front end. Values for cylinder- methods. Translated into FMEP, this deviation leads to an error in specific torsion can be acquired using measuring technology, the range of 25 to 30 %. including the possibility of optical measurement of gas exchange TDC during fired engine operation. 11 Crankshaft Torsion and Energy Balance As shown in Figure 10, it can be seen from the comparison of the If validated results for cylinder-specific crankshaft torsion are energy balances determined that, even in this evaluation, the cor- available, then a corrected volume function for cylinder pressure rected methods exhibit a clear improvement in their average diag-nostics can be calculated. In this, the volume function values (cylinder averages). The result of a virtually closed energy (depending on crank angle) for each individual cylinder is calcu- balance is very promising. A study of the individual results of the lated in the maximum expression and used in the subsequent method “Fully corrected volume function" and "Cylinder-specific engine process calculation. TDC correction" reveals a different, cylinder bank-specific behaviour. The energy balances in Bench 1 for the "Cylinder-

www.kistler.com 9 If individual cylinder torsion can be determined at TDC, then a 13 Publication references simplified correction method can be applied. This means that the angle of torsion can be defined directly in the combustion analy- [1] Davis R., Patterson G.: sis system, and corrected even as the measurement is being per- Cylinder pressure data quality checks and procedures to formed. maximize data accuracy. SAE Paper 2006-01-1346 A third possibility is provided by the thermodynamic TDC adaptation of engine process calculation. This requires excellent coordi- [2] Tazerout M., LeCorre O., Rousseau S.: nation of gas exchange simulation for the determination of the TDC determination in IC engines based on thermodynamic required charge mass at the individual cylinders. analysis of the temperature-entropy diagram. SAE Paper 1999-01-1489 Where torsional effects were not taken into account for each operating point, in the case investigated deviations of a magni- [3] Piraner I., Pflueger C., Bouthier O.: tude of around 30 % in FMEP and around 2 % in energy balance Cummins crankshaft and bearing analysis process. were determined. 2002, North American MDI User Conference

[4] AVL Product description Excite Power Unit, Dynamics and acoustics of power units and drivelines, 2013

[5] Ciecinski M., Dolt R.: Combustion analysis of a 2-cylinder engine using a 60-2 crank speed sensor. 10th International Symposium on Combustion Diagnostics, 2012

[6] Jauk T., Wimmer A.: Neues Mess-System zur Bestimmung des oberen Totpunktes in Verbrennungsmotoren. 12th Conference "Der Arbeitsprozess des Verbrennungsmo- tors"

[7] Fusshoeller H., Bargende M.: Verfahren zur OT Bestimmung. 14. Kistler Indizierexperten Forum, 2008

10 www.kistler.com www.kistler.com 11 Europe, Asia,Americas,Australia Offices in www.kistler.com Tel. 5222411 +41 Switzerland 8408 Winterthur Eulachstrasse 22 Kistler Group