filter media through the backward extrapolation of the over- all filtration resistance versus filtrate weight or filtration time curves over the steady stages operating ...
A
Applied
X. Cao: Filtration resistance during pressure filtration tests of liquid aluminium alloys
Xinjin Cao Department of Metallurgy and Materials, The University of Birmingham, Birmingham, England
© 2006 Carl Hanser Verlag, Munich, Germany
www.ijmr.de
Not for use in internet or intranet sites. Not for electronic distribution.
Filtration resistance during pressure filtration tests of liquid aluminium alloys Prefil Footprinter, a portable pressure filtration instrument, is usually employed to detect liquid metal quality. However, no investigations have ever been done to determine overall filtration resistance and resistance of the “clean” filter media. Based on the identification of flow behaviour using the derivative methods, classic filtration equations have been effectively used to estimate total filtration resistance for the first time. The variations of overall resistance during filtration provide a new approach to understand filtration behaviour. The resistance values of the “clean” filter media have been obtained for the first time. An innovative method has been developed to obtain the resistance of the “clean”' filter media through the backward extrapolation of the overall filtration resistance versus filtrate weight or filtration time curves over the steady stages operating in incompressible cake mode.
time graph with earlier benchmarks seems to provide a measure to assess the cleanliness of liquid metal. Examination of the material trapped in and above the filter can further confirm the results and extend the interpretation. Recently the filtrate weight versus filtration time curves have been analysed using 1st and 2nd derivative methods [8, 9]. It is interesting to find that there probably exist 3 phases during the pressure filtration tests. These are initial transient, steady, and terminal transient states. Based on the classification of the three stages, conventional filtration theories have been successfully used to understand the pressure filtration tests of liquid aluminium alloys [10 – 14]. In this work, the filtration resistance is investigated for the first time for the Prefil Footprinter tests of liquid aluminium alloys. To better understand the filtration behaviour, the classic filtration equations of cake mode are introduced in the following section.
Keywords: Pressure filtration; Prefil Footprinter; Al alloy; Filter medium; Resistance
2. Filtration equations of cake mode
1. Introduction Currently there are two quantitative methods to measure solid inclusion contents in molten aluminium alloys: electrical resistance and pressure filtration [1]. A Liquid Metal Cleanliness Analyser (LiMCA), based on the electrical resistance measurement, is thought to provide truly in-line detection of inclusion quantities and sizes, though the equipment is expensive, not portable, and requires a significant amount of maintenance [1 – 3]. The methods based on the pressure filtration or pre-concentration techniques include Porous Disc Filtration Analysis (PoDFA) from Alcan and Liquid Aluminium Inclusions Sampler (LAIS) from Union Carbide [3]. The metallurgical analyses of the concentrated samples can provide a qualitative and quantitative evaluation of the inclusions though these methods are very expensive, being both time-consuming and effectively offline procedures. In addition, experienced metallurgical technical personnel are required to carry out the inclusion analyses [2]. Another portable Pressure Filtration Technique (Prefil Footprinter) can provide both real-time (on-line) melt filtration curves and concentrated metallurgical samples for further inclusion evaluation [4 – 7]. The method, developed and patented by N-Tech, England, is based on the PoDFA technique combined with a real-time data acquisition system that allows the recording of the filtrate weight versus filtration time profile while the test is being performed. Comparison of the filtrate weight versus filtration Int. J. Mat. Res. (formerly Z. Metallkd.) 97 (2006) 8
Filtration is usually used to separate suspended solids from slurry (here liquid metal) through a filter medium. The pores of the filter medium are small enough to prevent the passage of some of the solid particles. As filtration proceeds, some solid material is left behind and forms a cake on top of the filter medium which serves as an additional inline filter as shown in Fig. 1 [15 – 18]. During the filtration test, the filter medium resists the passage of the fluid, under the influence of a force which is the pressure differential across the filter medium. The cake thickness increases from 0 to Lc, corresponding to an increase in filtrate volume collected from 0 to V. The flow velocity of the filtrate (u) is proportional to the pressure difference DP imposed over the cake and the filter medium; the overall fluid velocity is inversely proportional to the
Fig. 1. Schematic diagram showing the flow of liquid metal through the cake and filter medium during the pressure filtration test.
1163
A
Applied X. Cao: Filtration resistance during pressure filtration tests of liquid aluminium alloys
viscosity of the flowing fluid l and the resistance of the cake and the filter medium [19 – 21]:
© 2006 Carl Hanser Verlag, Munich, Germany
www.ijmr.de
Not for use in internet or intranet sites. Not for electronic distribution.
u¼
1 dV DP DP ¼ ¼ A dt lR lðRc þ Rm Þ
ð1Þ t¼
The filtration resistance results from the frictional drag on the liquid as it passes through the cake and filter medium [21]. Rm is the resistance of the filter medium (m – 1). If the cake is uniformly distributed on the filter surface and is incompressible, then the cake resistance Rc is directly proportional to the filtrate volume V and inversely proportional to the filter area A [21]: Rc ¼ r
V A
ð2Þ
Where = Specific cake resistance (m/kg), and r = Mass of solid cake captured per unit filtrate volume (kg/m3). Substituting Eq. (2) into Eq. (1) gives dV DP ¼ l V dt r þ Rm A A
ð3Þ
or dt lr lRm ¼ Vþ dV A2 ðDPÞ AðDPÞ
ð5Þ
W2 þ
lRm W qAðDPÞ
ð6Þ
The filtration process involves two principal resistances, i. e. the resistance of the filter cake Rc and the resistance of the filter medium Rm. The medium can be regarded as a barrier with an interconnected pore structure. The resistance of the filter medium (Rm) is the inverse (reciprocal) of the permeability, which is directly related to the porosity of the medium. Thus a high permeability is an indication of high porosity, low particle retentivity, and thereby low medium resistance [19]. Clearly, the resistance of the filter medium can be calculated if the intercepts of dt/dW versus W curves (Eq. (5)), or the first order factors of t versus W curves (Eq. (6)) are known. These resistance values are termed as effective medium resistance which is usually higher than the resistance of the “clean” filter medium due to the deposition of fine particles in the pores of the filter media [12, 22]. To obtain the resistance of the “clean” filter medium, one classic approach is to extrapolate backwards R – V curves to zero volume [18, 23]. Another method is to apply the Darcy’s law to the medium [19, 24]: Rm ¼
Equation (5) also yields a linear function between the inverse mass flow rate (dt/dW) and the filtrate weight (W). A straight line should be expected between dt/dW and W if the filtration operates in an incompressible cake mode. lr lRm and the intercept can be obThe slope 2 2 q A ðDPÞ qAðDPÞ tained from the plot. During the course of filtration the pressure and flow rate may change or remain constant. Thus the mass filtration rate dW/dt can be integrated for various modes of filtration 1164
lr 2q2 A2 ðDPÞ
ð4Þ
The specific cake resistance (m/kg) is introduced as a measure of metal flow resistance. It represents the flow resistance to filtration for a unit mass of solid cake per unit filter area [21]. This parameter is known to be a function of the alloy system, degree and type of processing, the filter medium itself, and the type of inclusions being filtered [17]. Under constant pressure conditions for a given melt this parameter () is considered to be invariant. The resistance of the cake increases with time as the cake builds up (i. e. it becomes harder to force the filtrate through the cake, as the cake becomes thicker). Thus, filtration becomes slower and slower, i. e. the filtrate volume rate progressively decreases as filtration progresses under constant pressure conditions. Eq. (4) yields a linear function between the inverse volumetric flow rate dt/dV and filtrate volume V [15 – 21]. To determine the specific cake resistance, Eckert et al. [17] investigated the linear relationships between dt/dV versus V for 1100, 5082, and 7075 aluminum alloys. Now filtrate volume is replaced by filtrate weight W (g) and the density of the filtrate q (kg/m3) is assumed to be constant. Then we have dt lr lRm ¼ Wþ dW q2 A2 ðDPÞ qAðDPÞ
such as constant pressure, or constant mass flow rate. If a constant pressure drop (DP) across the cake and filter is assumed, and considering that at time t = 0, W = 0. Thus,
AðDPm Þ qAðDPÞ ¼ dV dW jt ¼ 0 l l dt dt
ð7Þ
At the initial stage of filtration before any particle deposition has occurred, the pressure loss across the filter medium is equal to the overall filtration pressure loss [18, 24]. So the resistance of the “clean” filter medium can be determined as long as the volume or weight of filtrate collected versus filtration time curve has been measured.
Nomenclature A = Cross-sectional area of the cake and filter (m2), Lc = Thickness of cake (m), Lm = Thickness of filter medium (m), P = Pressure (Pa), DP = Pressure difference or drop across the cake and filter medium (Pa), Rc = Resistance of filter cake (m – 1), Rm = Resistance of filter medium (m – 1), R = Total filtration resistance (m – 1), V = Volume of filtrate that has passed the filter during time t (m3), W = Weight of filtrate that has passed the filter during time t (kg), q = dV/dt = Volumetric filtration rate (m3/s), t = Filtration time (s), u = Filtrate velocity (m/s), q = Density of filtrate (kg/m3), qs = Density of solid inclusions (kg/m3), = Specific cake resistance (m/kg), l = Viscosity of filtrate (Pa × s or N-s/m2), and r = Solid inclusion mass captured per unit filtrate volume (kg/m3). Int. J. Mat. Res. (formerly Z. Metallkd.) 97 (2006) 8
A
Applied
X. Cao: Filtration resistance during pressure filtration tests of liquid aluminium alloys
© 2006 Carl Hanser Verlag, Munich, Germany
www.ijmr.de
Not for use in internet or intranet sites. Not for electronic distribution.
Table 1. Alloy compositions and experimental details used in the work. No.
Alloy compositions
Precipitation and sedimentation processing & melting temperatures
Temperatures at filtration
1 2 3 4 5 6 7 8 9 10 11 12 13
Al-11.52Si-0.38Mg-0.57Fe-0.59Mn-0.17Ti Al-11.52Si-0.38Mg-0.57Fe-0.59Mn-0.17Ti Al-11.52Si-0.38Mg-0.57Fe-0.59Mn-0.17Ti Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.52Si-0.38Mg-0.57Fe-0.59Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-11.12Si-0.36Mg-1.14Fe-1.09Mn Al-7.5Si-0.4Mg-0.5Fe-0.3Mn-0.13Ti Al-7Si-0.4Mg (commercial UK 2L99)
No processing; melted at 730 °C 600 °C for 10 min, reheated to 730 °C 600 °C for 10 min, reheated to 730 °C No processing; melted at 760 °C No processing; melted at 760 °C 600 °C for 10 min, reheated to 730 °C 600 °C for 30 min, reheated to 760 °C 600 °C for 60 min, reheated to 760 °C 600 °C for 120 min, reheated to 760 °C 585 °C for 120 min, reheated to 760 °C 600 °C for 240 min, reheated to 760 °C No processing; melted at 760 °C No processing; melted at 730 °C
700 °C 700 °C 700 °C 700 °C 720 °C 700 °C 720 °C 720 °C 720 °C 700 °C 720 °C 720 °C 700 °C
Notes: Alloy compositions given in wt.% unless noted otherwise. The holding temperature for alloy 10 was not well controlled, leading to the occurrence of Al solidification during the precipitation and sedimentation processing. After the processing the top clean liquid was decanted from the bottom precipitate-rich material except for Trial 3. Trials 2 and 3 were reheated in an electric resistance furnace. All other trials were reheated in an induction furnace after the decantation.
From Eq. (1) the total filtration resistance can also be calculated: R ¼ Rm þ Rc ¼
AðDPÞ qAðDPÞ ¼ dV dW l l dt dt
ð8Þ
From the above equations, it is evident that filtration resistance and resistance of the “clean” filter medium are important parameters for understanding the mechanisms of the pressure filtration tests. The variations of the overall resistance during the tests may provide a new approach to understand the filtration behaviour. The medium resistance can be used to calculate the flow resistance, flow velocity, pressure distribution and loss, drag forces and to characterise medium quality. Thus, the overall filtration resistance and medium resistance have significant influences on the structure of the consequently formed cake and the filtration behaviour. It is important to have quantitative information on these resistance values in order to better understand the filtration process. However, no investigations have ever been done to determine the filtration resistance and the resistance of the “clean” filter medium for the Prefil Footprinter tests of liquid aluminum alloys. This work attempts to estimate the overall filtration resistance and the resistance of the “clean” filter medium, which will provide a useful basis to better understand the filtration mechanisms.
α-Al(MnFe)Si particles and solid inclusions. After the melt has been held for the preset sedimentation time, approximately 2.2 kg out of 3 kg of the sedimented melt was carefully separated from the precipitate-rich part at the base of the crucible. The purified melt (approximately 2.2 kg) was reheated to 730 or 760 °C and then about 2 kg metal was used to conduct the Prefil Footprinter test. The crucible was made of a low heat capacity, highly insulating fibrous material equipped with a filter of an average pore size of 90 lm. When the metal cooled to either 700 or 720 °C, the pressure chamber was closed and pressure was applied to the liquid metal in the crucible, pushing it through the filter. As filtered metal fell into the weigh ladle it was weighed every 3 seconds. After the pre-set test time (150 seconds) the pressure was automatically released from the pressure chamber and the filtration test was terminated. During the test of the filtration, two stage pressures were employed. A high-pressure of about 0.21 MPa (30 psi) was used to initiate flow through the filter. When the filtrate weight of more than 20 grams was recorded at the load cell the pressure was switched to a low value of approximately 0.08 MPa (12 psi) to maintain metal flow through the filter for the remainder of the filtration time [28]. Trials 1, 4, 5, 12, and 13 were not subjected to precipitation and sedimentation processing and were directly evaluated using the Prefil Footprinter instrument.
4. Results and discussion 3. Experimental procedures 4.1. Filtration resistance The experimental details were given in some earlier publications [8 – 14] and therefore are only briefly described here. The materials used were nominal cast Al-7/11.5Si0.4Mg alloys (British designations LM9, LM25 and 2L99) with various Fe, Mn, and Ti contents (Table 1). A mass of 3 kg of the experimental alloy was melted at 730 or 760 °C in an induction furnace. Some alloys were subjected to precipitation and sedimentation processing at various temperatures and periods of time [25 – 27]. The molten metal was cooled to 585 or 600 °C and held in a holding furnace for 10 to 240 minutes to sediment primary Int. J. Mat. Res. (formerly Z. Metallkd.) 97 (2006) 8
A filtrate weight versus filtration time curve was obtained after each Prefil Footprinter test. The first derivative method was used to analyse the flow behaviour during the filtration tests. Three stages, i. e. initial transient, steady, and terminal transient stages were identified for all experiments as shown in Table 2 [9] in which the compressibility of the cakes formed on the filter media has been investigated based on the linear relationship between inverse mass flow rate (dt/dW) and filtrate weight (W) (Eq. 5) over the steady stages [11]. Total filtration resistance is calculated using 1165
A
Applied X. Cao: Filtration resistance during pressure filtration tests of liquid aluminium alloys
© 2006 Carl Hanser Verlag, Munich, Germany
www.ijmr.de
Not for use in internet or intranet sites. Not for electronic distribution.
Table 2. Three stages determined using the 1st derivative method and cake filtration mode for steady stages [9, 11]. No.
Filtrate weight (g)
Initial transient stage (s)
Steady stage (s)
Terminal transient stage (s)
Cake mode for steady stage
1 10 12 2 5 3 9 8 7 6 4 13 11
201 288 291 400 454 540 605 785 788 981 994 1130 1305
0 – 12 0–6 0 – 24 0 – 33 0–6 0 – 150 0 – 138