Professional Documents
Culture Documents
Fuel
journal homepage: www.elsevier.com/locate/fuel
Combustion Division, Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg, Sweden
Keywords: A study was conducted to determine how geometry-induced cavitation affects the reliability of injection rate
Nozzle geometry estimates based on momentum flux measurements. Experiments were performed with two hydro-grinding diesel
Cavitation injector nozzles (Do = 190 µm), one having a cylindrical nozzle orifice that has a strong tendency to induce
Momentum flux cavitation, and the other having a convergent nozzle orifice that tends to suppress cavitation. In addition, near-
Injection rate
field microscopic images were captured to investigate the influence of cavitation on the liquid jet at the nozzle
Near-field jets
exit. The momentum flux measurements and near-field imaging were conducted at ambient pressure, using
injection pressures of up to 2500 bar. When estimating the injection rate based on momentum flux measure-
ments, failure to account for cavitation caused the integrated injection rate to be overestimated around 9% for
the cavitating (cylindrical) nozzle but not for the non-cavitating (convergent) nozzle. We therefore recommend
that cavitation should be accounted for when estimating injection rates for cavitating nozzles but not for non-
cavitating nozzles. Further, we show that the tendency of a given nozzle orifice to induce cavitation can be
evaluated by measuring the nozzle’s momentum flux under the assumption of zero cavitation and then com-
paring the estimated mass of injected fuel to the weighed mass.
1. Introduction similar choking of the mass flow through a cavitating nozzle under
diesel engine-like conditions was observed by Payri et al. [12]. Sangiah
Internal combustion (IC) engines are likely to remain the dominant et al. [8] and Emberson et al. [10] used hydro-grinding (or positive
means of vehicular propulsion for the foreseeable future [1]. The radius of inlet) nozzles with positive K-factor (K = (Di Do)/10 µm
combustion process in conventional diesel engines is controlled by the where Di and Do are the inlet and outlet diameter of the nozzle orifice).
quality of fuel-air mixing, which strongly affects both engine-out The measurement of fuel injection rate may be appropriate with such
emissions and engine efficiency. Evaluation of injector performance by nozzles, because positive K-factors or hydro-grinding inlets are known
measuring the fuel mass flow rate or the momentum flux of the jet to suppress cavitation. Ganippa et al. [7] and Luo et al. [9] did not
provides useful information that can be used to better control the in- provide the K-factor and hydro-grinding of nozzle orifices, and they
jection process, and thus fuel-air mixing quality. There are various applied the approach that disregards the effect of cavitation. This would
methods for measuring the fuel injection rate, such as the Bosch method be questionable for the cavitating nozzles. However, some approaches
[2], the Zeuch method [3] and a method based on momentum flux that account for reductions in the effective orifice area due to cavitation
measurements [4]. The latter method has the advantage of simplicity bubbles or foaming in the nozzle orifice have also been used [4,6]. To
and has therefore been adopted by several research groups [5–10]. our knowledge, the former and the latter methods have never been
However, there are two different approaches to measure the fuel in- compared using experimental data for nozzles with cavitating and non-
jection rate based on momentum flux measurements. The approach cavitating geometries.
based on momentum flux measurements that do not account for the The objective of this paper is thus to explore the influence of nozzle
occurrence of cavitation in the nozzle orifice have been used [5,7–10], geometry-induced cavitation on estimates of fuel injection rates based
despite the possibility that they may yield inaccurate results in cases on momentum flux measurements. To this end, experiments were per-
where cavitation occurs and has significant effects on the mass flow. It formed with two hydro-grinding nozzles – one having a cylindrical
has been reported that cavitation generates foams that adhere to the orifice with a strong tendency to induce cavitation, and the other
nozzle orifice and choke the mass flow through the nozzle [11]. A having a convergent orifice that tends to suppress cavitation. In
⁎
Corresponding author.
E-mail addresses: chengjun.du@chalmers.se (C. Du), sven.b.andersson@chalmers.se (S. Andersson), f3cma@chalmers.se (M. Andersson).
https://doi.org/10.1016/j.fuel.2018.10.119
Received 12 June 2017; Received in revised form 18 October 2018; Accepted 23 October 2018
Available online 30 October 2018
0016-2361/ © 2018 Elsevier Ltd. All rights reserved.
C. Du et al. Fuel 238 (2019) 354–362
Fig. 1. Simplified cross-sectional view of flow in the cavitating and the non-cavitating nozzle orifices. Ao : the area of the orifice exit, Aeff : the cross-sectional area of
flow, : the density of flow, f : the liquid density, u : the flow velocity profile, ueff : the effective velocity profile of the flow through the effective area Aeff .
2. Theoretical background Combining Eqs. (2) and (3), the actual mass flow rate of liquid jet
can be expressed as:
As discussed in the ‘Introduction’ section, there are two different mact = f ·A eff ·M (4)
methods to estimating injection rates based on momentum flux mea-
surements. The main difference between these methods is whether they From Bernoulli’s equation, the theoretical velocity uth , i.e. the
do or do not account for the effects of cavitation in the nozzle orifice. maximum potential fluid velocity at the orifice exit, can be expressed
Cavitation creates vapor bubbles or foam films along the inner walls of as:
the nozzle orifice. Fig. 1 presents simplified depictions of flow phe-
nomena in the cavitating nozzle orifice and the non-cavitating nozzle 2·(Pi Pa)
uth =
orifice. Fig. 1(a) shows that the cavitating layer may contain vapor f (5)
bubbles or foam films in a cavitating nozzle orifice. Naber and Siebers
[4] took account for the effect of cavitation by assuming that it reduces The theoretical mass flow rate mth can be expressed as:
the cross-sectional area of flow ( Aeff ) such that it becomes smaller than mth = f · A o ·uth = Ao · 2· f ·(Pi Pa) (6)
the area of the orifice exit ( Ao ). The effective velocity profile (ueff ) of
the flow through the effective area Aeff is then computed, taking into where Pi and Pa represent the injection pressure and ambient pressure,
account the flow’s non-uniform velocity profile (u ). It should be noted respectively. Ao represents the theoretical cross-sectional area of the
that cavitation introduces variation in the density of the flow at the liquid jet at the nozzle exit if there is no cavitation and the jet’s cross
orifice exit because of the presence of vapor bubbles; for simplicity, it is sectional area is equal to that of the nozzle exit. The discharge coeffi-
assumed that the density of flow ( ) through the area Aeff is equal to the cient Cd is defined as the ratio of the actual mass flow rate to the the-
liquid density ( f ). The alternative approach, which has been widely oretical mass flow rate, and is obtained by combining Eqs. (4) and (6):
used [5,7–10], ignores the effects of cavitation and thus effectively
assumes that Aeff Ao . This approach may be appropriate for a non- mact Ca·M
Cd = =
cavitating nozzle orifice such as a hydro-grinding nozzle with a positive mth 2· Ao ·(Pi Pa) (7)
K-factor because nozzle orifices with positive K-factors or hydro-
The area contraction coefficient Ca and velocity coefficient Cv are
grinding inlets are known to suppress cavitation. Specifically, Blessing
defined as done by Payri et al. [12]:
et al. [13] and Duke et al. [14] found that nozzle orifices of the size
used in diesel injectors induced no cavitation if they were hydroground Aeff
Ca =
and had positive K-factors. This raises the question of whether the ap- Ao (8)
proach that ignores cavitation can produce reliable results for cavi-
ueff
tating nozzles, and if not, why? The purpose of this paper is to address Cv =
uth (9)
these issues.
The derivation of the approach that accounts for cavitation in- The coefficient Ca is used to account for the reduction in the liquid
corporates that of the approach that disregards cavitation. If the liquid flow area through the orifice exit because of cavitation bubbles. Under
jet’s velocity in the radial direction is neglected, its momentum flux cavitating conditions, both Cv and Ca are less than 1. The relationship
when it impinges on the target’s surface normal to the direction of li- between Cd , Ca and Cv is defined as:
quid flow can be expressed as:
Cd = Ca· Cv (10)
M = mact · ueff (1) Combining Eqs. (7) and (8), Ca can be derived as:
where mact and ueff represent the actual mass flow rate and the effective 2· Ao ·Cd2·(Pi Pa)
Ca =
velocity of the liquid jet, respectively. The actual mass flow rate mact M (11)
can be expressed as:
The only unknown parameter in the above equation is Cd , which can
mact = f ·A eff ·ueff
be determined using an expression presented by Naber and Siebers [4]:
(2)
m
where f and Aeff represent the fuel density and the effective area of the Cd =
t· Ao · f ·uth (12)
liquid jet at the orifice exit, respectively.
Combining Eqs. (1) and (2), the effective velocity of liquid jet can be where m is the mass of fuel injected over a time t . Having de-
expressed as: termined Cd , Ca can be computed. This approach was applied by Siebers
355
C. Du et al. Fuel 238 (2019) 354–362
[15], who used the calculated value of Ca with a scaling law to predict 3. Experimental set-up
liquid phase penetration. If assuming Ca 1, one obtains the definition
of the discharge coefficient Cd for the approach that neglects cavitation: 3.1. Set-up for momentum flux measurement
Pressure sensor
Accumulator
Injector
NI chassis Pulse generator driver
356
C. Du et al. Fuel 238 (2019) 354–362
5.25
10.50
15.75
K0 K2
21.00
Fig. 3. Demonstration of the liquid jet impingement. The green line is the
boundary of the liquid jet. Injection pressure is 2000 bar. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web
version of this article.)
357
C. Du et al. Fuel 238 (2019) 354–362
~920 mm
(a)
Injection
period
Fig. 8. Comparison of the original injection pressure data (black line) and
processed one (red line). The condition is the same as in Fig. 7. (For inter-
pretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
paper.
(b)
5. Results and discussions
358
C. Du et al. Fuel 238 (2019) 354–362
K0 K2
=9.1
=1.9
=8.6 =8.4 =1.6
=9.3 =1.7
=-1.0
=9.2
=-1.9
Fig. 9. Weighed fuel masses m w (filled black circles) and masses calculated based on Cd (mc , open circles) for nozzles K0 and K2. The double standard deviations (2σ)
of the calculated mc values and the differences are also shown (filled blue circles). (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
359
C. Du et al. Fuel 238 (2019) 354–362
K0 K2
2500 bar
2500 bar 2000 bar
2000 bar 1600 bar
1600 bar 1200 bar
1200 bar
800 bar
800 bar
Fig. 12. Transient injection rates at different injection pressures for nozzles K0 (left) and K2 (right).
with the value of 0.83 estimated in the section on ‘Results based on the effective area of the liquid jet to be largely independent of the injection
calculation of Cd ’, and implies that when using Eq. (13), the main error pressure. Accordingly, the discharge coefficient for nozzle K0 could be
derives from overestimation of Ca . also largely independent of the injection pressure.
Finally, attempts were made to measure the transient mass flow
rates for both nozzles using Eq. (4). For this purpose, the area con-
traction coefficient Ca was assumed to be constant, yielding the tran- 5.4. Near field liquid jet imaging at the nozzle orifice exit
sient mass flow rates shown in Fig. 12. Unsurprisingly, the use of Eq. (4)
means that the trends in the mass flow rate mirror those in the mo- As discussed above, the difference in the mass flow rate between the
mentum flux. As shown in Fig. 12, the mass flow rate for nozzle K2 is nozzle K0 and the nozzle K2 is due to cavitation inside the former
higher than that for K0, presumably because of the former nozzle’s nozzle. To better understand the effect of cavitation on the liquid jet, it
higher discharge coefficient. The difference in the injection duration at was imaged near the nozzle orifice exit using a long-distance micro-
different injection pressures in Fig. 12 was due to different hydraulic scope during the steady period of injection (1.5–3.0 ms ASOE). As
forces for closing injector needle at different injection pressures. shown in Fig. 7(a), the momentum flux was relatively stable during this
period, so it can be assumed that the injector needle was sufficiently
raised to avoid affecting the results. Thus any observable differences in
5.3. Examination of cavitation in the nozzle K0 liquid jet structure between the two nozzles are due to differences in the
nature of their orifices. Fig. 13 shows instantaneous liquid jet images at
Eq. (15) can be used to evaluate the extent of cavitation in the cy- 2.0 ms ASOE. The black line shown in Fig. 13 is the boundary of the
lindrical nozzle K0. Payri et al. [17] concluded that the critical cavi- liquid jet, and the threshold for determining the liquid jet boundary is
tation number CNcri was 1.2 for an injection pressure of 800 bar, so the same for both nozzles. The difference in light intensity between the
cavitation should be expected when the CN is below 1.2. They also nozzles is due to the use of different light source intensities and does not
found that the critical cavitation number CNcri increased with the in- affect the determination of the jet boundary.
jection pressure. The calculated cavitation number CN for the nozzle K0 The mean width of the jet at different distances from the nozzle tip
is 1.0 at all injection pressures, so it is expected to exhibit cavitation is shown Fig. 14. The mean jet width was obtained by averaging the
independently of the injection pressure. Cavitation in nozzle K0 causes liquid jet width during the steady period (1.5–3.0 ms ASOE) over five
the effective area of liquid flow to be smaller than that of the orifice injections. Since the discharge coefficient Cd is relatively stable at dif-
exit, so the area contraction coefficient Ca is less than 1. This is con- ferent injection pressures for both nozzles, the width is only shown for
sistent with the finding that the calculated mass for this nozzle is injection pressures of 2000 and 2500 bar. The width measurements
greater than the weighed mass, as shown in Fig. 9. The mass flow rate
and discharge coefficient Cd calculated under the assumption of negli-
gible cavitation are thus not reliable for cavitating nozzles such as K0, K0
so there will be no further discussion based on this value. Instead, the
discussion for this nozzle will use the discharge coefficient Cd calculated
using Eq. (12), and the associated area contraction parameter.
Payri et al. [17] and Soteriou et al. [11] found that the discharge
coefficient only depends on the cavitation number CN under cavitating
conditions. In addition, Payri et al. found that the choking of mass flow
rate under cavitating conditions increased with the injection pressure.
However, the data presented in Fig. 12 do not indicate that nozzle K0 K2
exhibited such a choking. This difference may be due to the use of
different ambient pressures and a different cylindrical nozzle in this
work and that of Payri et al.; we used a cylindrical nozzle orifice with
30% inlet hydro-grinding, while Payri et al. used one with a much
sharper inlet which would induce stronger cavitation in a nozzle. On
the other hand, using a transparent nozzle under diesel-like injection
conditions, Badock et al. [21] observed that increasing the injection
pressure had little effect on the dimensions of the cavitation films or the Fig. 13. Instantaneous near field liquid jet images for nozzles K0 (top) and K2
accumulation of bubbles. In other words, one would thus expect the (bottom). Conditions: injection pressure 2000 bar, 2.0 ms ASOE.
360
C. Du et al. Fuel 238 (2019) 354–362
50 µm 50 µm
Fig. 14. Liquid jet width near the orifice exit. Conditions: injection pressure 2000 bar (left), injection pressure 2500 bar (right). Double standard deviations ( ± 2σ)
are shown for the nozzle K0.
361
C. Du et al. Fuel 238 (2019) 354–362
assumption of no cavitation and comparing the result to the actual Alfredsson and Kim Kylström from Scania CV AB are also acknowledged
injected mass. It should be noted that the assumption of no cavitation in for taking X-ray images and assembling the injectors. Electron micro-
this paper is made when the measured area contraction coefficient Ca is scope images were taken by Philipp Hoier at Department of Industrial
approximately 1. However, a small amount of cavitation, not sig- and Materials Science, Chalmers University of Technology.
nificantly altering the mass flow rate, may still be present, and to fur-
ther confirm absence of cavitation in nozzle K2, an independent in- References
vestigation, for example, X-ray based imaging might be helpful.
It should be noted that the method of measuring transient mass flow [1] Kalghatgi GT. The outlook for fuels for internal combustion engines. Int J Engine
rates based on momentum flux measurements has some limitations. As Res 2014. https://doi.org/10.1177/1468087414526189.
[2] Bosch W. The fuel rate indicator: a new measuring instrument for display of the
discussed in the section on “Results based on the calculation of Cd ”, this characteristics of individual injection. SAE Technical Paper 660749 1966. https://
method assumes that the discharge coefficient Cd is constant during the doi.org/10.4271/660749.
injection period. However, it is known that the Cd is not constant. [3] Postrioti L, Buitoni G, Pesce FC, Ciaravino C. Zeuch method-based injection rate
analysis of a common-rail system operated with advanced injection strategies. Fuel
Therefore, mass flow rate estimates based on momentum flux mea- July 2014;128(15):188–98. https://doi.org/10.1016/j.fuel.2014.03.006.
surements do not account for the impact of needle lift on the transient [4] Naber J, Siebers D. Effects of gas density and vaporization on penetration and
mass flow rate. It was concluded by Manin et al. [24] that relatively dispersion of diesel sprays. SAE Technical Paper 960034 1996. https://doi.org/10.
4271/960034.
high oscillations of mass flow rate were mainly due to displacement of
[5] Johnson J, Naber J, Tang M, Taylor Z, et al. Investigation of the impact of im-
the injector itself while injecting. Therefore, in cases where it is ne- pingement distance on momentum flux rate of injection measurements of a diesel
cessary to account for temporal variation in the mass flow rate, the injector. SAE Technical Paper 2015-01-0933 2015. https://doi.org/10.4271/2015-
01-0933.
Bosch long-tube method or Zeuch method are recommended. Although
[6] Pickett L, Manin J, Payri R, Bardi M, et al. Transient rate of injection effects on spray
there are limitations on the momentum flux measurement method, one development. SAE Technical Paper 2013-24-0001 2013. https://doi.org/10.4271/
of advantages is that it is possible to measure the mass flow rate of each 2013-24-0001.
nozzle hole in a multi-hole injector. [7] Ganippa L, Andersson S, Chomiak J. Transient measurements of discharge coeffi-
cients of diesel nozzles. SAE Technical Paper 2000–01-2788 2000. https://doi.org/
10.4271/2000-01-2788.
7. Conclusion [8] Sangiah DK, Ganippa LC. Application of spray impingement technique for char-
acterisation of high pressure sprays from multi-hole diesel nozzles. Int J Therm Sci
2010;49(2):409–17. https://doi.org/10.1016/j.ijthermalsci.2009.08.001.
Using two 30% hydro-grinding nozzles with a K-factor of 0 and K- [9] Luo F, Cui H, Dong S. Transient measuring method for injection rate of each nozzle
factor of 2, two different mass flow rate measurement approaches using hole based on spray momentum flux. Fuel 2014;125:20–9. doi:0.1016/
the momentum flux measurement have been discussed. Liquid jet j.fuel.2014.02.011.
[10] Emberson DR, Ihracska B, Imran S, Diez A, Lancaster M, Korakianitis T. Hydraulic
images near the nozzle tip were also captured using a long-distance characterization of Diesel and water emulsions using momentum flux. Fuel
microscope. The following conclusions were drawn: 2015;162:23–33. https://doi.org/10.1016/j.fuel.2015.08.016.
[11] Soteriou C, Andrews R, Smith M. Direct injection diesel sprays and the effect of
• For a cavitating nozzle K0, the measured area contraction coefficient cavitation and hydraulic flip on atomization. SAE technical paper 950080 1995.
https://doi.org/10.4271/950080.
Ca is around 0.85 which is significantly less than 1. When estimating [12] Payri R, Garcia JM, Salvador FJ, Gimeno J. Using spray momentum flux mea-
the injection rate based on momentum flux measurements, failure to surements to understand the influence of diesel nozzle geometry on spray char-
acteristics. Fuel 2005;84(5):551–61. https://doi.org/10.1016/j.fuel.2004.10.009.
account for cavitation caused the integrated injection rate to be
[13] Blessing M, König G, Krüger C, Michels U, et al. Analysis of flow and cavitation
overestimated around 9%. Thus, it is recommended that for mass phenomena in diesel injection nozzles and its effects on spray and mixture forma-
flow rate estimates based on momentum flux measurements, the tion. SAE Technical Paper 2003-01-1358 2003. https://doi.org/10.4271/2003-01-
approach that accounts for cavitation by separately determining the 1358.
[14] Duke D, Swantek A, Tilocco Z, Kastengren A, et al. X-ray imaging of cavitation in
discharge coefficient should be applied. diesel injectors. SAE Int J Engines 2014;7(2):1003–16. https://doi.org/10.4271/
• For a non-cavitating nozzle K2, the measured area contraction 2014-01-1404.
[15] Siebers D. Scaling liquid-phase fuel penetration in diesel sprays based on mixing-
coefficient was around 1 because there was little or no cavitation
limited vaporization. SAE Technical Paper 1999–01-0528 1999. https://doi.org/10.
expected in the nozzle orifice. When estimating the injection rate 4271/1999-01-0528.
based on momentum flux measurements, failure to account for ca- [16] Nurick WH. Orifice cavitation and its effects on spray mixing. J. Fluids Eng.
vitation caused less than 2% the difference between the integrated 1976;98(4):681–7. https://doi.org/10.1115/1.3448452.
[17] Payri F, Bermudez V, Payri R, Salvador FJ. The influence of cavitation on the in-
injection rate and the actual weighed mass. Therefore, the approach ternal flow and the spray characteristics in diesel injection nozzles. Fuel
that disregards cavitation is appropriate for such nozzles. 2004;83(4–5):419–31. https://doi.org/10.1016/j.fuel.2003.09.010.
• The tendency towards cavitation in a nozzle orifice can be estimated [18] Chaves H, Knapp M, Kubitzek A, Obermeier F, et al. Experimental study of cavi-
tation in the nozzle hole of diesel injectors using transparent nozzles. SAE Technical
by comparing the injected mass calculated using the approach that Paper 950290 1995. https://doi.org/10.4271/950290.
neglects cavitation to the actual weighed mass: if the nozzle pro- [19] Payri R, Salvador FJ, Gimeno J, Bracho G. The effect of temperature and pressure
duces appreciable cavitation, the calculated mass will be over- on thermodynamic properties of diesel and biodiesel fuels. Fuel
2011;90(3):1172–80. https://doi.org/10.1016/j.fuel.2010.11.015.
estimated.
•
[20] Huang W, Moon S, Ohsawa K. Near-nozzle dynamics of diesel spray under varied
Because of cavitation in nozzle K0, the width of the liquid jet near needle lifts and its prediction using analytical model. Fuel 2016;180:292–300.
this nozzle’s tip was greater than that for the non-cavitating nozzle https://doi.org/10.1016/j.fuel.2016.04.042.
K2, in keeping with others’ findings. Both nozzles produced an intact [21] Badock C, Wirth R, Fath A, Leipertz A. Investigation of cavitation in real size diesel
injection nozzles. Int J Heat Fluid Flow 1999;20(5):538–44. https://doi.org/10.
liquid core at and near the orifice exit. 1016/S0142-727X(99)00043-0.
• It was also found that for both nozzles K0 and K2, the discharge [22] Du C, Andersson M, Andersson S. Effects of nozzle geometry on the characteristics
of an evaporating diesel spray. SAE Int J Fuels Lubr 2016;9(3):493–513. https://
coefficient and area coefficient were independent of the injection
doi.org/10.4271/2016-01-2197.
pressure in the 800–2500 bar range. [23] Liu Z, Im K, Wang Y, Fezzaa K, et al. Near-nozzle structure of diesel sprays affected
by internal geometry of injector nozzle: visualized by single-shot X-ray imaging.
Acknowledgments SAE Technical Paper 2010–01-0877 2010. https://doi.org/10.4271/2010-01-0877.
[24] Manin J, Kastengren A, Payri R. Understanding the acoustic oscillations observed in
the injection rate of a common-rail direct injection diesel injector. J Eng Gas
Financial support from CERC (Combustion Engine Research Centre Turbines Power 2012;134(12):122801https://doi.org/10.1115/1.4007276.
at Chalmers) and its participant companies is gratefully acknowledged,
especially Scania CV AB who provided injection systems. Sara
362