Experimental Characterization and Modeling of High Performance

Article pubs.acs.org/IECR

Experimental Characterization and Modeling of High Performance Structured Packings† Ž arko Olujić,*,‡ Thomas Rietfort,§ Helmut Jansen,§ Björn Kaibel,§ Egon Zich,§ Günther Frey,§ Gerhard Ruffert,⊥ and Torsten Zielke⊥ ‡

Process & Energy Department, TU Delft, Leeghwaterstraat 44, NL-2628 CA Delft, The Netherlands Julius Montz GmbH, Hofstrasse 82, 47003 Hilden, Germany ⊥ Bayer Technology Services GmbH, Building B 310, 51368 Leverkusen, Germany §

ABSTRACT: Results are presented of the total reflux distillation experiments carried out with the newest generation of J. Montz corrugated sheet structured packings, including metal sheet packings (B1-250MN, B1-350MN, B1-500MN), expanded metal packings (BS-500MN), and wire gauze packings (A3-500M), using a 0.59 m i.d. column and chlorobenzene/ethylbenzene as a test system. It appears that by combining appropriately the dimensions and design of corrugations and the corrugation inclination angle, either efficiency or capacity, or occasionally even both, can be enhanced significantly, depending on the type of the packing. This allows leaner columns in new designs and more capacity and other benefits in retrofit situations. The Delft model, with appropriate enhancement of the turbulent vapor flow mass transfer coefficient and the flow direction change related pressure drop, proved to be capable of approaching closely measured efficiency and pressure drop both qualitatively and quantitatively. The exception in this respect is the pressure drop of B1-500MN, with a significant underprediction that is more pronounced at 1 bar than at 0.1 bar, while the trend as well as the predicted points of onset of loading are close to observed ones.

1. INTRODUCTION As observed during hydraulic tests performed with an air/water system and a prototype of B1-250 M packing, a long smooth bend on the bottom side of corrugations proved to be a simple but highly effective geometry modification that enabled significant capacity increase with respect to conventional Montz B1-250.1 Indeed enabling a smooth transition of ascending vapor and, more importantly, for the descending liquid, which tends to accumulate at transitions of packing elements or layers at sufficiently high vapor load,2 was the key to success in this case. Subsequently the prototypes of B1-250 M and B1-350 M have undergone total reflux distillation tests, which were performed at the Separations Research Program (SRP) of the University of Texas at Austin in a column with the internal diameter of 0.43 m with cyclohexane/n-heptane (C6/C7) as test system at pressures of 0.17, 0.33, 1.03, and 4.14 bar, respectively.3 These tests confirmed that the M-series packings can be operated within an extended loading region at the same efficiency as conventional counterparts. In the meantime, a comparison of liquid distribution characteristics of B1-250 and B1-250M was carried out using the TU Delft 1.4 m diameter air/water column hydraulics simulator. The large scale liquid distribution and mixing, which occurs at transitions between packing elements, appeared to be consistently good and, in this respect, the performance of B1-250M was even slightly better.4 Interestingly, this packing exhibited a more pronounced wall flow, which however is avoided in practice by installing properly working wall wipers. Although the B1-250M exhibits a significantly lower pressure drop, the gas distribution performance appeared to be similar to that of the conventional counterpart and in both cases a consistently good one.5 In the meantime both B1-250M and B1-350M have been installed in © 2012 American Chemical Society

numerous industrial columns and proved to work accordingly in both new designs and revamps. However the above-mentioned total reflux distillation tests performed at SRP indicated a certain loss of efficiency in preloading range, where M-packings operate at a relatively lower pressure drop. The main part of the reduction in overall pressure drop comes from avoiding sharp bends at transitions of packing elements or layers. Another significant contribution in this respect results from the fact that a long bend reduces strongly the number of crossings of gas flow channels, which is reflected in a less pronounced pressure drop due to interaction of crossing vapor flows. The latter is generally considered to be potentially beneficial to mass transfer. First, intensive mixing of crossing gas flows assures maintenance of concentration gradient within a packing element, and second, it superimposes a swirling flow pattern forcing vapor to flow through the Vshaped channel at increased effective velocity, which should lead to increased skin friction, i.e. to a stronger interfacial interaction of two phases. Since capacity and efficiency are generally interdependent, and capacity is more affected by the liquid handling than by the pressure drop associated with transitions between subsequent packing elements, it was conjectured that reducing the length of a bend, i.e. increasing the number of gas/liquid interactions per flow channel length, could be beneficial for efficiency. A similar effect could be expected from a reduction in the corrugation inclination angle, i.e. going below the commonly used 45°, which however, according to the Delft model,6,7 leads to a Received: Revised: Accepted: Published: 4414

November 10, 2011 February 9, 2012 February 24, 2012 February 24, 2012 dx.doi.org/10.1021/ie202585t | Ind. Eng. Chem. Res. 2012, 51, 4414−4423

Industrial & Engineering Chemistry Research

Article

2. EXPERIMENTAL Figure 3 shows a flowsheet of the test facility available at Bayer Technical Services (BTS) in Leverkusen, Germany, used in this

significant increase in the pressure loss. So the challenge was to arrive at a favorable corrugation geometry, i.e. that maximizing the desired gain in efficiency with an acceptable penalty on the pressure drop side and vice versa, including standard sheet metal and expanded metal packings with a large specific geometric area. Figure 1 shows schematically the original structure and geometry parameters manipulated. Since the imperforated

Figure 1. Schematic illustration of the structure of “M” series packings, indicating manipulated geometry parameters.

surface of large specific geometric area packing B1-500 proved to be sensitive to upsets in efficiency at column loads associated with those achievable with high-capacity packings, the new so-called MN series packings have been provided with a regular pattern of holes (see Figure 2). In addition to this,

Figure 3. Flowsheet of the test installation used in this study, with main equipment dimensions and utilities indicated.

study, including relevant equipment size and utilities related information. The heart of this facility is a 0.59 m internal diameter column that can accommodate packed beds up to 4 m height. The packings tested were three common sizes of new corrugated metal sheet B1-MN series packings, with specific geometric areas of 250, 350, and 500 m2/m3, respectively. Besides B1-series packings, also the new designs of expanded metal packings (BS-500MN) and wire gauze packings (A3500M) with specific geometric area of 500 m2/m3 were tested. Nominal specific geometric areas as well as the element and bed heights of the packings tested are shown in Table 1. Packings have been delivered in single elements and installed in heights either of ∼2 m (10 elements or layers) or ∼3.5 m (18−21 elements or layers), each packing layer rotated to the previous one by 90°. The test system was the ideal, close boiling mixture of chlorobenzene/ethylbenzene (CB/EB), which is also the standard system adopted long ago at Sulzer,8 and the operating pressures were 100 and 1000 mbar, respectively. The volume of initial charge of liquid, usually a mixture containing 25−40 mol %

Figure 2. Corrugated sheets of B1-MN type packing, with and without perforations.

these packings are characterized by a lower corrugation inclination angle and a shorter bend with respect to that of the M series, with a radius depending on the corrugation base dimension. Upon completing a series of screening tests at TU Delft, packing geometry adjustments (perforations, corrugation bend length, and inclination angle balanced) have been implemented and representatives of different types and sizes of new packings thoroughly tested using state of the art total reflux distillation facilities available at Bayer Technology Services in Leverkusen, Germany. These data have been used to validate, modify accordingly, and improve the predictive accuracy of the Delft model.7 4415

dx.doi.org/10.1021/ie202585t | Ind. Eng. Chem. Res. 2012, 51, 4414−4423


Article

Table 1. Dimensions of Packings Considered in the Present Study specific geometric area, m2/m3 packing element height, m packed bed height, m

A3-500

A3-500M

B1-250MN

B1-350MN

B1-500MN

BS-500MN

BS-500MN (no holes)

500 0.166 3.49

500 0.169 3.55

250 0.20 2.00

350 0.20 2.00

500 0.194 3.49

500 0.191 1.91

500 0.191 1.91

chlorobenzene, was large enough to avoid any component depleting effect. The vapor is generated in a falling film evaporator and the column load is related to reboiler load, which is arranged by adjusting the inlet pressure and flow of steam (maximum 15 bar). The overhead vapor enters a water cooled condenser and an after-cooler, connected in series. The latter is used to control the extent of reflux subcooling. To ensure the proper initial liquid distribution, for each of two operating pressures, a separate narrow trough liquid distributor was used. For 1 bar operation, a large turndown distributor covering a range of specific liquid loads from 3.3 to 33 m3/m2·h (or m/h) was used. This distributor employed 56 equidistantly arranged drip tubes, which is an equivalent to 205 drip points per square meter of column cross sectional area. One should note that this irrigation density was considered appropriate to cover the needs of testing packings with specific geometric areas ranging from 250 to 500 m2/m3. Owing to the fact that specific liquid loads as encountered at the low end of F-factors in the case of 0.1 bar operation were well below the minimum value of the large turndown distributor, the packings with the largest specific geometric area were tested under vacuum conditions using a Montz type S distributor covering the specific liquid load range from 0.37 to 7.3 m3/m2·h. This distributor employed 20 drip tubes equipped with spider-like extensions containing seven short dripping legs, increasing the number of drip points to 140, which is equivalent to 506 drip points per square meter of column cross sectional area. Such an irrigation density ensures proper initial wetting of a packing surface even at specific liquid loads well below 1 m3/m2·h. As illustrated in Figure 1, a vane (chevron) type liquid collector is installed below the bed to collect the liquid raining down from the bed and deliver it via a down-pipe to the bottom of the column. The purpose of this device is twofold, i.e. avoiding interference of ascending vapor and draining liquid, and removing entrained liquid from vapor leaving the reboiler, which could be expected at higher vapor loads, particularly under vacuum conditions. The test series started usually with running the column under maximum reboiler load (flooding conditions) to ensure complete wetting of packing before the start of measurements. Usually 12 points were measured, and for each of them, 3 liquid samples were taken in regular time intervals. Top samples were taken from the reflux line, and the bottom samples were taken from the feed line to the falling film reboiler. The first sample was taken 8 h after reaching steady state, i.e. upon complete stabilization of pressure, temperature, and flow profiles. Subsequent samples were taken with 1 h in between. Liquid samples were analyzed using a precalibrated gas chromatograph. Reflux (liquid) flow rate was measured using a coriolisflowmeter, and was used as the basis for determination of column load, i.e. the F-factor, which, in turn, was determined using mass flow rate of vapor at the top, corrected for the effect of subcooled reflux. Pressure drop was measured directly using a pressure difference cell. Since the distance between pressure taps above and bellow the bed was larger than the bed height,

measured pressure drop was corrected for static vapor column height contribution. Pressure and the temperature in the top of the column, the temperature of the reflux and the temperature in the sump of the column were continuously measured and recorded. Upon completion of a test series, measurements were repeated for three different column loads, to check reproducibility, which appeared to be very high throughout this study. Relevant properties of the mixture, i.e. densities, viscosities, specific heat capacity, enthalpy of evaporation, and vapor pressure were determined on additive basis for top and bottom conditions. Geometric average was used for column relative volatility, which together with measured top and bottom conditions was used in conjunction with the Fenske equation to determine the number of theoretical plates, N, contained in the bed. Owing to the fact that bottom sample was taken from the liquid circulation line of the reboiler, the number of calculated theoretical plates (equilibrium stages) was reduced by one, i.e. that equivalent to separation performance of the partial reboiler as employed in present case. One should note that depending on the operating conditions and the configuration employed the separation performace of partial reboiler may vary to some extent. However by taking the contribution of partial reboiler to be equivalent to one equilibrium stage, a conservative estimate of packed bed efficiency is assured. Due to a reduced total number of equilibrium stages involved, the potential negative effect is pronounced in the case of shorter beds employed in the present study (∼2 m). This is compensated to a certain extent by the fact that shorter beds tend to exhibit better efficiency relative to longer beds (∼3.5 m). The latter are long enough to allow liquid maldistribution to develop in lower part of the bed, affecting adversely the overall efficiency. However, both effects are limited and resulting efficiencies can be considered as representative for the performance of packings considered under given test conditions. Both, efficiency expressed as the bed height equivalent to a theoretical plate [HETP = hbed/(N − 1)] and measured bed pressure drop were plotted as a function of the average F-factor, i.e. the vapor load of the column. The latter is an arithmetic average of top and bottom values. One should note that in total reflux experiments (L = G) the specific liquid load increases proportionally to the increasing vapor load, to the extent depending on the ratio of liquid and vapor densities, i.e. operating pressure. As shown in Table 2, which contains representative values of relevant physical properties at middle of the bed conditions, going from 0.1 to 1 bar operating pressure means a factor three increase in the specific liquid load. Another important physical property affected strongly in present case is the vapor diffusion coefficient that changes nearly proportionally to the pressure ratio. Also, one should note that the test system employed represents a close boiling mixture, with a rather low relative volatility, which as well as the corresponding slope of the equilibrium line remains nearly constant along the bed. According to Ottenbacher et al,9 such a low relative volatility, only slightly affected by a 10-fold change in pressure, makes CB/EB system suitable for measurements 4416



Article

similar to that of B1-250MN, which reaches the maximum efficient capacity at a pressure drop of around 9 mbar/m and performs still well under hydraulic flood conditions. Interestingly, at the lowest F-factor the efficiencies of three packings are close to each other, and with increasing F-factor tend to diverge, approaching at the high end the values corresponding approximately to the ratios of specific geometric areas. This indicates indirectly that at lower loads the low area packing uses the installed area much more effectively than the packing with largest specific geometric area. This is clearly illustrated in Figure 5, which shows the relative surface utilization efficiency, expressed by dimensionless

Table 2. Physical Properties (Average at Middle of the Bed Conditions) of Chlorobenzene/Ethylbenzene System at 0.1 and 1 bar, Respectively property/pressure (bar)

0.1

1

temperature, °C liquid density, kg/m3 liquid viscosity, Pa s liquid diffusivity, m2/s vapor density, kg/m3 vapor viscosity, Pa s vapor diffusivity, m2/s surface tension, N/m relative volatility, − slope of equilibrium line, − reference composition (x), − liquid load, m3/m2·h, at F-factor = 2 m/s·(kg/m3)0.5 or Pa0.5

67 930 5 × 10−4 3.4 × 10−9 0.409 8 × 10−6 4 × 10−5 0.025 1.18 0.99 0.5 4.95

134 870 3 × 10−4 6.4 × 10−9 3.233 1 × 10−5 4.2 × 10−6 0.020 1.13 1.00 0.5 14.9

including high number of equilibrium stages, without compromising accuracy. A detailed overview of equations including relevant parameters and mixing rules that allow determination of physical properties of the CB/EB system with utmost rigor can be found elsewhere.9

3. RESULTS AND DISCUSSION 3.1. Experimental Evidence. Figure 4 shows total reflux distillation performances of B1-250MN, B1-350MN, and Figure 5. Relative packing surface utilization efficiency as a function of F-factor.

product of specific geometric area and corresponding HETP values, as a function of F-factor. The lower the value of the product of the specific geometric area (ap) and the measured HETP, the better the packing performance. Ideally, different packing sizes should exhibit the same value, which however is here only partly the case, i.e. with B1-250MN and B1-350MN at the higher end of F-factor values. This dimensionless parameter, expressed as the ratio of the specific geometric area and the number of theoretical plates per meter bed height, has been introduced under the name “surface efficiency” by Fischer et al.10 Data published in their paper also indicate that the wetted fraction of the installed surface area of a corrugated sheet structured packings tends to decrease with increasing specific geometric area. According to the situation shown in Figure 5, the best utilization of the installed area is evident in case of B1-250MN, and this is most pronounced at lowest F-factors, where this packing exhibited approximately 1.6 times better efficiency than at the loading point. One should note that Sulzer data from measurements carried out with the Mellapak 250.Y and Mellapak 252.Y packings with the same system but in larger diameter columns, e.g. 1 and 1.2 m do not exhibit any efficiency enhancement at low F-factors.11 Although the column diameter could play a role here, it appears that this is a distinctive characteristic of B1-series packings, resulting from a specific interaction between corrugation dimensions (macro), the shape of the corrugation ridge (rounded or sharp), and packing surface texture (micro geometry). Such a high efficiency of B1-250MN at the lowest F-factor is surprising to some extent, because the specific liquid load is at its low end and in the case of 0.1 bar operation the related

Figure 4. Measured efficiency, pressure drop, and capacity of the B1250MN, B1-350MN, and B1-500MN as a function of F-factor at 0.1 bar, including B1-500MN data obtained at 1 bar.

B1-500MN packings as observed at 0.1 bar, including for comparison the data for B1-500MN obtained at the pressure of 1 bar. As expected both the efficiency and pressure drop increase, while the capacity decreases with increasing specific geometric area. One should note that the observed efficiency for this close boiling system is generally high but tends to decrease in more or less pronounced way with increasing F-factor, depending on the size of packing specific geometric area. Accordingly, 250 m2/m3 packing is more sensitive than 350 m2/m3 packing, while the 500 m2/m3 packing seems to be insensitive in this respect. Striking is the performance of B1-350MN; it approaches the larger specific area packing in efficiency and smaller specific area packing in capacity. Indeed, it appears that this fully optimized packing can operate smoothly at very high capacity, 4417



Article

absolute values are very low, i.e. 1−3 m3/m2·h. Since in this region the mass transfer kinetics is at its lowest level, which is to a limited extent counterbalanced by long vapor and liquid contact times, it seems that this particular corrugation size and shape stimulates active wetting of a relatively larger fraction of installed area than is the case with larger area packings B1350MN and B1-500MN. This is in accordance with earlier experiences gained with conventional B1 and B1-M series packings,3 indicating that this is something inherent to the geometry of shallow embossed surface corrugated sheets where the radius of the crimp (fold) angle tends to decrease with increasing specific geometric area, i.e. decreasing dimensions of corrugations. Namely, a factor of 2 difference in the specific geometric area implies the same difference in the number of parallel flow channels. Doubling the number of flow channels in conjunction with sharper corrugation ridges means a reduced capability of the liquid to go over the corrugation ridges to neighboring channels, thus less effective wetting of the installed surface area. An increased liquid load should be beneficial in this respect and should be reflected in an increased packing efficiency. However, as shown in Figure 4, the efficiency of B1-500MN exhibits the same trend and same values at respectively 0.1 and 1 bar operating pressure, while in the latter case the specific liquid load is approximately three times larger. Identical behavior has been observed with Sulzer packings tested with the same system at similar operating pressures.11 The absence of efficiency improvement may mean that the beneficial effect of increased effective area may be in this case counterbalanced by an adverse pressure effect on the mass transfer kinetics side (much lower values of vapor diffusion coefficient!). As expected, on the hydraulics’ side, increasing the operating pressure leads to an increased pressure drop and correspondingly reduced packing capacity. Unlike common B1-series packings, the B1-MN series packings compared in Figure 4 are provided with perforations, i.e. a regular pattern of 4 mm holes. These appeared to be essential only for B1-500 packing to perform accordingly in loading region. However, as shown in Figure 6, the expanded metal packing with the same area and corrugation design (BS500MN) does not require perforations to perform well over the whole range of operating conditions. According to the measured efficiency curves shown in Figure 6, imperforated

BS-500MN packing exhibits a smooth and stable performance up to the point of onset of flooding, which is in the latter case some 15% beyond that of perforated counterpart as well as B1500MN. Interestingly, in the preloading region, the perforated packing, with a regular pattern of 4 mm holes, which effectively reduce the installed area up to 10%, exhibits nearly equivalent loss of efficiency, which however disappears upon reaching the point of onset of loading, where both perforated and imperforated packing exhibit similar efficiency. As well-known, BS-series packings are made of expanded metal sheets, containing a large fraction of small rhombic perforations, which under loading conditions provide a good hold for liquid and exhibit an absorbing effect which allows the liquid to be pushed through to the other side of the sheet to allow pressure equalization of neighboring channels. This may be an explanation for stable operation well beyond the flooding point of B1-500MN as well as for a relatively lower pressure drop, because the surface saturated by liquid appears to be smoother and delivers less frictional resistance to vapor flow. A considerably higher price of the material makes expanded metal packings less interesting economically; however, the beneficial pressure drop at practically equal efficiency could make it to be an interesting alternative for B1-series packings in some applications. Pressure drop limited applications are traditionally the domain of more expensive, but in this respect, highly effective wire gauze packings as Sulzer BX and Montz A3. Figure 7

Figure 7. Measured efficiency, pressure drop, and capacity of respectively conventional wire gauze packings A3-500 and A3-500M as a function of F-factor at 0.1 bar.

shows the performance of well established gauze packing A3500 and its high capacity counterpart A3-500M (long bend in lower section of corrugated sheets), respectively. Both packings exhibit practically equal efficiency, which, as expected, tends to deteriorate steadily with increasing F-factor, and is significantly lower than that of B1-500MN at high F-factors. However, due to a corrugation inclination angle of 60° to horizontal, the pressure drop of A3-500 is 2−3 times lower than that of B1500MN in the range of commercial interest. According to Figure 7, A3-500 M exhibits even a significantly lower pressure drop but this does not bring any capacity gain compared to conventional A3-500 packing. Anyhow, such an extremely low pressure drop is a great advantage in pressure drop sensitive applications, which means that this packing could be used to revamp columns equipped with conventional gauze packings, where appropriate.

Figure 6. Measured efficiency, pressure drop, and capacity of respectively perforated and imperforated expanded metal packing BS-500MN as a function of F-factor at 0.1 bar. 4418



Article

3.2. Delft Model Validation. 3.2.1. Mass Transfer Efficiency. For B1-series packings and operating conditions evaluated in this study, Figure 8 shows the predicted ratios of

transfer coefficient may be considered as a wrongdoer in the present case. According to DM,6,7,12 and to cover the whole range of operating conditions encountered in typical total reflux distillation tests, the vapor side mass transfer coefficient, kG (m/s), comprises both a laminar and a turbulent part and is described as kG =

k G,lam 2 + k G,turb2

(1)

with ⎛D ⎞ d k G,lam = ⎜ G ⎟0.664ScG1/3 ReGrv hG d l ⎝ hG ⎠ G,pe

(2)

and ⎛D ⎞ k G,turb = ⎜ G ⎟ ⎝ dhG ⎠ Figure 8. Predicted effective area for the packings and conditions shown in Figure 4.

ξ φ ReGrv ScG GL 8

1 + 12.7

⎡ ⎛ d ⎞2/3⎤⎥ ⎢ hG ⎟ ⎢1 + ⎜⎜ l ⎟ ⎥ ⎝ G,pe ⎠ ⎥⎦ ⎢⎣

effective and installed area, the latter being taken as 10% lower than the nominal one due to loss of area occupied by perforations. As expected the effective area increases with increasing F-factor, i.e. specific liquid load to the extent as described by the well-known Onda et al. correlation adapted to structured packings.1,6 The trend is the same but absolute values are highest for B1-250MN and tend to decrease with increasing specific geometric area to the extent imposed by the corresponding empirical correction term adopted recently in the Delft model.7 According to the model, the effect of the operating pressure, which in the present case increases the specific liquid load by a factor three (see the table inset in Figure 8) is strong, resulting in an effective area enhancement of respectively around 60% at the lower end and around 40% at the upper end of the F-factor. Figure 9 shows the predicted performance of the B1-250MN packing as a function of the F-factor, illustrating the effects of changes in governing variables. The Delft model (DM, large squares) overpredicts the measured efficiency strongly. The fact that even the calculated HETP curve based on installed area as effective area (ae = ap) still lies well above the measured one indicates that not the effective area but the vapor side mass

ξGLφ 8

(ScG2/3 − 1)

(3)

2

where DG (m /s) is the gas phase (vapor) diffusion coefficient, ScG (−) is the gas phase Schmidt number, ReGrv (−) is the gas phase Reynolds number based on relative velocity, dhG (m) is the hydraulic diameter of the gas flow (triangular) channel, lG,pe (m) is the length of the gas flow path within a packing element height, ξGL (−) is the gas−liquid interface friction factor, and φ (−) is the wall fraction of the triangular flow channel occupied by the liquid film, while subscripts lam and turb denote laminar and turbulent flow related vapor phase mass transfer coefficients, respectively. The V-shaped fraction of the cross section of triangular gas flow channel occupied by liquid film is described by φ=

2s b + 2s

(4)

where b (m) and s (m) represent corrugation base and side dimension, respectively. For a corrugated sheet structured packing with a fold or apex angle of 90°, the fraction of crosssectional area available for liquid is approximately 0.59, regardless the packing size, but it can fluctuate slightly depending on actual corrugation base and height dimensions. Indeed, the correlation for turbulent vapor side mass transfer coefficient accounts directly for the friction between two phases on the interface, which however is limited to V-shaped part, i.e. fraction of the triangular flow channel that is used by liquid. Therefore, at certain stage of Delft model development12 coefficient φ (phi) was introduced, which effectively reduces the estimated value to approximately 60% of that estimated by original correlation,13 anticipating that additional enhancement in mass transfer could result from gas−gas interaction at the open planes formed between crossing flow channels. Due to exchange of momentum, whose strength depends strongly on the corrugation inclination angle, as suggested in ref 14, a more or less pronounced swirling motion is imposed to vapor flow in each flow channel, which may change to some extent along the length of each flow channel in both intensity and pattern. In general, this means that effective velocity of vapor flow might

Figure 9. Efficiency, pressure drop, and capacity of B1-250MN at 0.1 bar: predicted vs measured. 4419



Article

be much higher than that based on the assumption of axial flow. Here a detailed computation fluid dynamics (CFD) modeling approach could be revealing; however it is difficult to isolate and quantify experimentally this complex flow mechanics situation in a way that could generate data needed for model validation. However this is beyond the scope of the present work, which simply goes back to the roots by assuming the validity of original correlation, i.e. eq 3 with φ = 1. Figure 10 illustrates the extent of change in values of the vapor phase mass transfer coefficient (kG), as calculated using

Figure 11. Efficiency, pressure drop, and capacity of the B1-350MN at 0.1 bar: predicted vs measured.

Figure 10. Predicted kG values for packings and conditions shown in Figure 4, indicating the effect of change in the value of φ (phi).

the DM with φ = 1, for all packings and conditions considered in this study. In all cases the kG increases gradually with increasing F-factor, with absolute values and steepness of the slope increasing accordingly with increasing specific geometric area of the packing. The kG-values based on φ = 1 are approximately factor 1.7 higher than those estimated using φ-values corresponding to the dimensions of V-shaped flow channel (see eq 4). As demonstrated in Figure 9, if we apply DM with φ = 1 in conjunction with the estimated effective area (large triangles), assuming that perforations reduce the installed by 10% (Ω = 0.1), a fairly good agreement is obtained between measurement and model, with predicted values on the safe side. Taking the nominal specific geometric area as installed area (small triangles) results in a roughly 10% improvement in efficiency, which however leads to a slight underprediction in the loading region. It appears that using the original turbulent vapor flow mass transfer coefficient correlation in the Delft model12 leads to sufficient enhancement on the mass transfer coefficient side to achieve a close approach to observed efficiency. Therefore, this simple and effective modification has been adopted in further work in conjunction with established DM correlations7 for effective area and liquid side mass transfer coefficient determination. Figure 11 shows comparison of observed and predicted HETP values for B1-350MN packing, indicating again a good agreement, with the predicted curve mainly on the safe side. Figures 12 and 13 show the comparison of predicted and measured efficiencies for B1-500MN packing for, respectively, 0.1 and 1 bar operating pressure. Interestingly, with φ = 1 DM prediction in case of 0.1 bar is close but on the optimistic side,

Figure 12. Efficiency, pressure drop, and capacity of the B1-500MN at 0.1 bar: predicted vs measured.

Figure 13. Efficiency, pressure drop, and capacity of the B1-500MN at 1 bar: predicted vs measured.

which may mean that predicted effective area is larger than the actual one, while in case of 1 bar the agreement between measured and predicted efficiency curves is nearly perfect. This means that the model nicely balances the efficiency loss associated with nearly a factor of ten lower value of diffusion coefficient of the vapor with the efficiency gain resulting from a quite large increase in effective area (see Figure 8) due to a factor of three increase in the specific liquid load. Figure 14 shows comparison of observed and calculated HETP values for imperforated and perforated expanded metal 4420



Article

Interestingly, by increasing the value of vapor side mass transfer coefficient, the height of vapor side mass transfer unit is reduced, leading effectively in present constellation to certain increase in the fraction of the liquid side resistance. The latter tends to increase with increasing F-factor, i.e. specific liquid load, from approximately 8% at low-side to 18% at high-side of the F-factor, which is doubling with respect to that according to original DM. This is significant and brings us to another issue, i.e. the relative magnitude of the liquid side resistance in this and similar distillation applications, which is addressed in greater detail elsewhere.15 3.2.2. Pressure Drop. As shown in Figure 9, original DM slightly underpredicts the measured pressure drop. However, if the number of flow directions is changed from one per packing layer to two, as really encountered in case of MN series packings, then overall pressure drop increases to the extent shown in Figure 9. With this, a nearly perfect match is realized, with the predicted values on the safe side. Therefore, this change has been adopted and used throughout this study, with exception of wire gauze packing A3-500M which employs a much longer (smoother) bend in lower part of corrugation than MN-series packings. According to the situation shown in Figure 11, the predicted pressure drop for B1-350MN corresponds to the measured one until the point of onset of loading, while in the loading region the predicted values increase more pronouncedly than the measured ones. Similar to the situation with B1-250MN, the predicted loading point matches the measured one. In case of B1-500MN, the model underpredicts the measured curve and this is more pronounced at higher operating pressure (see Figures 12 and 13), where the contribution of the static vapor pressure, which is significant at 1 bar, has been accounted for. In both cases, the predicted pressure drop increases within the loading region at a steeper angle than the measured one. The predicted loading points agree well with observed ones. The extent of underprediction in the case of B1-500MN packing is significant, which is surprising to some extent because the model reflects in a physically sound way the geometry and physical properties related changes. One should note that the observed increase in measured pressure drop is not proportional to the ratio of installed areas. It is generally larger, and in case of two extremes, i.e. B1-500MN and B1250MN (specific geometric area ratio = 2.07) the ratio of measured pressure drop at the same load (FG = 1) is 2.54, which is well above the predicted value (1.96). This suggests that, similar but opposite to the situation with the effective area and efficiency, the vapor while ascending through irrigated beds of a large specific geometric area B1-series packing experiences additional flow resistances, not recognized yet and consequently not accounted for properly in the model. A weakness revealed on this occasion is incapability of the DM to account properly for observed operating pressure effect. This is mainly due to insufficiency in that respect of the empirical correlation describing the wall zone pressure drop. All equations of the DM related to hydraulics of corrugated sheet structured packing, including where appropriate the flow direction change angle as encountered in high performance corrugated sheet structured packings, can be found in a most recent paper describing dimensioning of packed dividing wall columns.16 One should note that most recently all DM equations, as given in ref 6, have been reproduced in a paper by Rahimpour et al.,17 including an error in the expression describing the

packing BS-500MN, indicating that model exhibits a good trend but lies on the optimistic side, while the predicted effect

Figure 14. Efficiency, pressure drop, and capacity of respectively perforated and imperforated expanded metal packing BS-500MN at 0.1 bar: predicted vs measured.

of 10% difference in installed area agrees fairly well with the observed one, the latter however being limited to the preloading region only. Figure 15 shows comparison of observed and calculated HETP values for two versions of wire gauze packing A3-500. Model predictions follow the observed trends perfectly, calculated values being some 20% higher, i.e. on the safe side.

Figure 15. Efficiency, pressure drop and capacity of respectively conventional wire gauze packings A3-500 and A3-500 M at 0.1 bar: predicted vs measured.

Small difference in predicted efficiencies of two packings corresponds to the difference in actual installed areas. On the other hand, if we would here take installed nominal area (Ω = 0) as basis for estimation of effective area, predicted curves would perfectly match the measured ones. It is interesting to mention that DM without any specific, gauze packing related provisions ensures such a good approach to actual situation in this case. With this in mind, we may conclude that achieved agreement between measured and estimated efficiencies justifies using the original correlation for turbulent vapor flow mass transfer coefficient.13 Adopting correspondingly increased value of gas side mass transfer coefficient leads to a better balance between two governing variables, the other being the effective area, while the effect of the liquid side resistance remains limited. 4421



Article

exhibits a fairly good accuracy, the predictions may appear optimistic when design of industrial columns is considered. Anyhow, since it does not require any empirical packing specific constant, it can be used for performance screening or conceptual design purposes involving untested corrugated sheet structured packings.

fraction of gas flow channels (Ψ) ending at the column walls, i.e. Equation 43 in ref 6, which is correctly given as eq 22 in ref 15. Namely, by separating the column cross sectional area into bulk and wall zone, with the fraction of the latter depending on the corrugation inclination angle, DM accounts for the observed column diameter effect.18 On the other hand, in the case of expanded metal packing with the same area (BS-500MN; see Figure 14), predicted pressure drop agrees perfectly well with the measured one in the preloading region, exhibiting a steeper increase in the loading region. As mentioned before, the measured pressure drop of BS-500MN is significantly lower than that of the equivalent B1-series packing. As shown in Figure 15, DM underpredicts significantly the pressure drop of A3-500 and A3-500M, which however is still within reasonable limits bearing in the mind the fact that microand macrogeometry of wire gauze packings differs to some extent from that of corrugated metal sheet packings. Anyhow, the fact that the predicted effect of the long bend in bottom part of the corrugations is much lower than observed is another point of concern.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. † Based on a paper presented at the Distillation and Absorption 2010 Conference (September 12−15, 2010, Eindhoven, The Netherlands).

■ ■

ACKNOWLEDGMENTS On the occasion of the 100th Anniversary of the Julius Montz GmbH, Hilden, Germany.

4. CONCLUSIONS The test results indicate that dimensions and design of corrugations and the corrugation inclination angle can be arranged to affect the pressure drop in a way that will favor either efficiency or capacity, or even both, as it appeared to be the case with B1-350MN that exhibited the best overall performance characteristics. The strikingly good efficiency of B1-250MN exhibited at low vapor and liquid loads tends to deteriorate with increasing F-factor, while the efficiency of large specific geometric area sheet metal and expanded metal packings is only slightly affected by changes in F-factor. B1500MN requires a perforated area to perform accordingly and its efficiency appeared to be the same at 0.1 and 1 bar, while the imperforated BS-500MN packing exhibited at the same efficiency a significantly larger capacity than its perforated counterpart. This indicates that perforations in the surface (roughly 10% less installed area) do not affect the effective area within the range of normal operation. Hydraulically, all packings exhibited expected performance. The Delft model with corrected direction change related pressure drop appeared to be capable of approaching measured pressure drop very well, both in trend and absolute values. Only in the case of B1-500MN packing, the model exhibited underprediction, which is in the case of 1 bar operation very pronounced. Predictions of efficiency agree in most cases very well with observation, confirming that adopted enhancement of vapor side mass transfer coefficient is justified. Predicted effective area seems to be close to actual situation, and with the CB/EB system also, it appears that the area utilization efficiency under vacuum conditions tends to decrease with increasing specific geometric area of B1-series packings. However underprediction of efficiencies measured at 0.1 bar with B1-500MN and BS-500MN packings suggests that the extent of deterioration in effective area is somewhat larger than anticipated by DM. This as well as the failure to approach close enough the measured pressure drop of B1-500MN at 1 bar are points of concern and will be given adequate attention in the future model validation and improvement work. Although the Delft model predicts well the trends associated with changes in packing geometry and operating conditions and

NOMENCLATURE ae = effective (interfacial) area, m2 ap = specific geometric area, m2 b = corrugation base dimension, m DG = gas (vapor) phase diffusion coefficient, m2/s dhG = hydraulic diameter of gas flow (triangular) channel, m FG = F-factor or vapor load, Pa0.5 FG,lp = F-factor or vapor load at loading point, Pa0.5 G = molar flow rate of the vapor, kmol/h HETP = height equivalent to a theoretical plate, m hbed = packed bed height, m hbend = corrugation bend height, m hpe = packing element (layer) height, m kG = vapor phase mass transfer coefficient, m/s L = molar flow rate of the liquid, kmol/h lG,pe = length of gas flow path within a packing element height, m N = number of theoretical plates, − ReGrv = gas phase Reynolds number based on relative velocity, − ScG = gas phase Schmidt number, − s = corrugation side dimension, m uLs = specific liquid load, m3/m2·h x = mole fraction of more volatile component in liquid, −

Greek Symbols

Δp/Δz = pressure drop, mbar/m ξGL = gas−liquid interface friction factor, − φ = (phi) fraction of the triangular channel occupied by liquid, − Ψ = fraction of gas flow channels ending at the column walls, − Ω = (omega) fraction of packing surface occupied by holes, − Subcripts

■

calc = calculated exp = measured lam = laminar turb = turbulent

REFERENCES

(1) Olujić, Ž .; Jansen, H.; Kaibel, B.; Rietfort, T.; Zich, E. Stretching the capacity of structured packings. Ind. Eng. Chem. Res. 2001, 40, 6172−6180.

4422



Article

(2) Suess, Ph.; Spiegel, L. Holdup of Mellapak structured packings. Chem. Eng. Process. 1992, 31, 119−124. (3) Olujić, Ž .; Seibert, A. F.; Kaibel, B.; Jansen, H.; Rietfort, T.; Zich, E. Performance Characteristics of a New High Capacity Structured Packing. Chem. Eng. Proces. 2003, 42, 55−60. (4) Olujić, Ž .; van Baak, R.; Haaring, J. P.; Kaibel, B.; Jansen, H. Liquid distribution properties of conventional and high capacity structured packings. Chem. Eng. Res. Des. 2006, 84, 867−874. (5) Olujić, Ž . Comaprison of gas distribution properties of conventional and high capacity structured packings. Chinese J. Chem. Eng. 2011, 19 (5), 726−732. (6) Fair, J. R.; Seibert, A. F.; Behrens, M.; Saraber, P. P.; Olujić, Ž . Structured packing performance − Experimental evaluation of two predictive models. Ind. Eng. Chem. Res. 2000, 39, 1788−1796. (7) Olujić, Ž .; Behrens, M.; Colli, L.; Paglianti, A. Predicting the efficiency of corrugated sheet structured packings with large specific surface area. Chem. Biochem. Eng. Q. 2004, 18, 89−96. (8) Spiegel, L.; Meier, W. Distillation columns with the structured packings in the next Decade. Chem. Eng. Res. Des. 2003, 81, 39−47. (9) Ottenbacher, M.; Olujić, Ž .; Adrian, T.; Jödecke, M.; Groβmann, C. Structured packing efficiency − Vital information for chemical industry. Chem. Eng. Res. Des. 2011, 89, 1427−1433. (10) Fischer, L.; Bühlmann, U.; Melcher, R. Characterization of highperformance structured Packing. Chem. Eng. Res. Des. 2003, 81, 79−84. (11) Pilling, M.; Spiegel, L. Design characteristics and test validation of high performance structured packing. Separations Technology Topical Symposium, AIChE Annual Meeting, November 4−9, Reno, NV, USA; 2001; Vol. 1, pp 64−69. (12) Olujić, Ž .; Kamerbeek, A. B.; de Graauw, J. A corrugation geometry based model for efficiency of structured distillation packing. Chem. Eng. Process. 1999, 38, 683−695. (13) Olujić, Ž . Development of a complete simulation model for predicting the hydraulic and separation performance of distillation columns equipped with structured packings. Chem. Biochem. Eng. Q. 1997, 11, 31−46. (14) Henley, B.; Dunbobbin, B.; Bennett, D. A unified method for countercurrent vapor/liquid packed columns. 2. Equations for the mass-transfer coefficients, mass-transfer area, the HETP, and the dynamic holdup. Ind. Eng. Chem. Res. 1994, 33, 1222−1230. (15) Olujić, Ž .; Seibert, A. F. Predicting the liquid side mass-transfer resistance of structured packings: State of the art, Proceedings of Distillation 2011: The James R. Fair Heritage Distillation Symposium, AIChE Spring National Meeting, March 13−17, Chicago, Illinois, USA, 2011; pp 263−282. (16) Dejanović, I.; Mastijašević, Lj.; Jansen, H.; Olujić, Ž . Designing a packed dividing wall column for an aromatics processing plant. Ind. Eng. Chem. Res. 2011, 50, 5680−5692. (17) Rahimpour, M. R.; Momeni, H.; Paymooni, K.; Kiani, A. A new model for the calculation of height equivalent to theoretical plate in high pressure columns equipped with structured packings for iC4/nC4 separation. Ind. Eng. Chem. Res. 2011, DOI: org/10.1021/ie101856x. (18) Olujić, Ž . Effect of column diameter on pressure drop of a corrugated sheet structured Packing. Chem. Eng. Res. Des. 1999, 77, 505−510.

4423


Experimental Characterization and Modeling of High Performance

Recommend Documents