Quantification of re-evaporated mass from loaded fibre-mist eliminators† Burkhard Riss*ab Ewald Wahlmu�llera and Wilhelm Ho�flingerb aPROFACTOR GmbH,Wehrgrabengasse 1–5, A-4400 Steyr, Austria. E-mail: briss@pr-steyr.ac.at bInstitute for Chemical Engineering, Fuel and Environmental Technology, Vienna University of Technology, Getreidemarkt 9/159; A-1060 Wien, Austria. E-mail: whoeflin@fbch.tuwien.ac.at Received 7th April 1999, Accepted 21st June 1999 Airborne lubricant emissions are a serious health hazard for employees in the metal working industry.The basic components of lubricants are oils and additives for adapting the properties to achieve the process demands. The oils used in lubricants are either mineral, synthetic or of biological origin. The lubricants are used as water-emulsions and also as straight oils.Extreme process conditions cause considerable amounts of aerosol and vapour emissions of lubricants into the working environment. Fibre filters are used in industrial demisters for pollution control and also for sampling purposes. Re-evaporation of separated lubricants from loaded fibre filters causes increased vapour emissions.Quantification of lubricant vapour emission was the subject of this research. An apparatus and an appropriate procedure for measuring the dynamic behaviour of evaporative losses from fibre filters were developed. The test piece of a loaded fibre filter was fixed in a sampling probe according to VDI 2066. The vapour concentration of organic compounds in downstream air was measured in real-time by using a photo-ionisation detector (PID).The PID was checked by sampling the vapour on an oleophilic adsorbent resin, solvent extraction and quantitative IR analysis and also gravimetrically. The two basic processes of filtering volatile aerosols are, on the one hand, collecting droplets on the fibres and, on the other hand, the evaporation of collected liquid. These two processes had to be separated in order to measure the increase of the vapour concentration caused by the tested fibre filter.The experiments were carried out using pure dodecane and hexadecane in order to avoid diYculties due to the unknown chemical composition of lubricant vapour. The variation of the air flow and the initial liquid mass on the filter covered the relevant range for industrial fibre demisters and for sampling methods based on collecting aerosols on fibre filters. It was found that the downstream air was saturated with lubricant vapour for a wide range of filter loads and filtration velocities.From the results obtained it can be concluded that loaded industrial filter systems emit air with saturation vapour concentration throughout their operation.Hence, vapour emissions can be estimated easily if the saturation vapour pressure and the mean relative molecular mass are known. Moreover, results obtained from measurements of filters with low loads confirm that the phase transition from liquid to vapour is influenced by the identified process parameters. This is relevant for measurement methods using fibre filters for sampling volatile aerosols.A semi-empirical model to estimate the evaporative losses for low loadings is proposed. However, further research is necessary to determine the full range of parameters that are relevant for evaporative losses from filters loaded with low masses of aerosols. of occupational health risks regarding aerosols and vapour is Introduction still under discussion.1,3,4 The assessment of occupational Extreme process conditions cause airborne emissions of metal health risks caused by airborne MWF emissions is either based working fluids (MWFs) at metal machining sites.Firstly, on the sum of aerosols and vapour (German approach: peripheral speeds of workpieces and cutting tools tear parts 10 mg m-3)1 or based on aerosols solely (Belgium, Finland of the MWFs into aerosols and fine droplets. Secondly, energy and UK: 5 mg m-3; Sweden: 3 mg m-3).In contrast, the dissipation at cutting processes and hot surfaces occurring at Austrian limit values at the workplace are 1 mg m-3 for thermal treatment applications cause considerable amounts of aerosols and 20 mg m-3 for the sum of aerosols and vapour.4 evaporated MWFs.1 A wide range of facilities such as electrostatic filters, sinter MWFs are applied either as oil-in-water emulsions or as plate element filters and centrifugal force filters are available straight oils.The basic ingredients of the lubricants are mineral to control airborne lubricant emissions at metal working sites. or synthetic oils and hydrocarbons of biological origin. Filter demisters are preferred frequently due to their high Additives adjust the properties (e.g.VT-index, corrosion, process reliability. The fact that loaded fibre mist collectors themselves can be a source of MWF vapour emissions was inhibiting the growth of micro-organisms) of the basic compothe subject of a recent investigation.5. nents to the specific process demands.2 In spite of the low Sampling lubricant aerosols on fine glass-fibre filters for vapour pressure of MWF components (‘semi-volatility’), conmeasurement purposes is common practice,6 although evapor- siderable amounts of vapour phase arise at ambient conditions.ative losses of collected aerosols during sampling may occur Exposure to airborne MWF emissions is known to be a due to the volatility of the MWFs.This results in systematic serious health hazard for employees.1 The current assessment underestimation of the aerosol phase. Recent measurements performed with diVerent sampling rates and equal sampling duration prove this assumption.7 †Presented at AIRMON ’99, Geilo, Norway, February 10–14, 1999. J. Environ. Monit., 1999, 1, 373–377 373Based on the results published in ref. 5 and 7, the goals of mass, the latter are proportional to the surface. The experiments were performed at loadings of the test piece below the this work are the identification and quantification of the most important parameters for evaporative losses from fibre filters saturation point of the filter material. The investigations should cover pure evaporative eVects, but exclude secondary emissions used for separation of liquid aerosol emissions.Therefore, an appropriate experimental set-up and a procedure to measure of liquid particles. Downstream aerosol loads indicated in Fig. 1(a) represent the dynamic behaviour of the evaporation process have to be developed. Losses of retained aerosols from fibre filters should penetrated and re-emitted particles.If the equilibrium between the liquid and vapour phase downstream of the filter is not be quantified. The variation of the selected parameters should include the range, which is important for industrial fibre filter achieved, evaporation still takes place if aerosols are present. Additional problems for in situ detection of the dynamic systems and for sampling devices.The results obtained are discussed in detail. behaviour of the evaporation process are caused by retention and distortion of the concentration front due to diVusion in the pipes and tubes between the filter and the measuring point. Experimental These eVects may not be negligible, taking account of the low air velocities present in sampling probes or highly eYcient Filtering volatile aerosols by using a fibre filter can be divided into two basic processes: collection of droplets on the filter industrial fibre filters.Penetration of aerosols was avoided by separating the filter loading process from the evaporation and evaporation of collected liquid. Evaporation takes place if the vapour pressure of the collected substance is below the process as mentioned above.Re-emission of aerosols downstream was neglected. Retention within the tube and lag time equilibrium (saturation vapour pressure) and liquid phase is still present. Evaporation of liquid induced by the filter takes of the detector were minimised by placing the suction inlet of the detector instrument as near as possible behind the test place if the vapour load downstream is higher than that upstream [Fig. 1(a)]. filter (approximately 5 cm). Because of the restricted suction city of the detector instrument with integrated pump, it The fundamental problem of measuring the concentration and ratio of liquid and vapour phase emissions by applying a was necessary to change the usual filtration process from pulling air through the filter to pushing it through by placing measurement method based on sampling the aerosols on a fibre filter (and collecting the vapour phase e.g.on an adsorbent the pump upstream as shown in Fig. 2. Flow conditions are regarded to be equal in these two operation modes. Pressure resin) is that the method itself causes a systematic error due to unknown evaporative losses of liquid from the filter during level diVers by about 1% (Dp#10 hPa) from absolute atmospheric pressure, so that changes of physical properties were sampling time.In order to avoid the diYculties in determining the ratio of aerosol and vapour load, the two basic processes regarded as negligible. The described phenomenon which had to be considered for were separated within the experiments. Therefore, the test filter was drizzled manually in calm air with liquid before separating volatile aerosols on fibre filters resulted in the experimental set-up given in Fig. 2. It consists of a standardised exposing it to the air flow. As given in Fig. 1(a), drain-oV mass flow of retained liquids filter holder according to VDI 2066,8 wherein a test filter of 50 mm diameter was implemented. The clearance of the filter may occur if the filter material becomes saturated itself. Saturation of filters implemented in a vertical position is given fixing was 40 mm.The variable air flow was provided by a gas sampler typically used for sampling applications (GS 312, by the balance of gravitational forces causing the tendency of drain-oV and forces caused by adherence of liquid to the fibre Desaga, Germany).The sampling device recorded automatically ambient air conditions and total air volume pumped surface. The former forces are proportional to the retained through the test filter. The pressure drop at the filter was detected by using a sensor based on a DMS bridge (Testo, Germany) and on-line recording (Testo 454, Testo). The dynamic behaviour of the downstream vapour concentration was measured with a portable photo-ionization detector (PID 2020, Perkin-Elmer, Canada).The PID works at 10.6 eV ionization radiation; thus, only organic air pollutants are detected. The permanent air gases (argon, carbon dioxide, nitrogen, oxygen, water vapour, etc.) require a higher energy of ionization and are not ionised by the UV photons. The PID enables in situ measurements of the downstream vapour concentration.The relevant parameters of the experiments are test filter properties, the range of air velocities penetrating the test filter, the upstream vapour load, the liquid load of the fibre filter, the Fig. 1 Model for filtration of mixed phase emissions (schematic). (a) Loaded air flow upstream: in industrial applications, liquid drain-oV is possible. (b) Situation of experiments with oil-wetted filter: vapour concentration downstream is filter induced only.SVC: saturation Fig. 2 Experimental arrangement (schematic). vapour concentration. 374 J. Environ. Monit., 1999, 1, 373–377ambient temperature, and the physical properties and chemical method for measuring hazardous substances, BIA 8000 (mineral oil, vapours and aerosols).6 Before starting the experimen- composition of the test fluid.Filter materials utilized for high performance filters in industrial systems applied for separating tal procedure, the PID was calibrated in a saturated atmosphere of the investigated liquid by placing the suction fine particles and those used for sampling are similar concerning the filter fibre diameter.The experiments were carried out inlet of the PID in a vapour saturated environment. Clean laboratory air for zero point adjustment was taken. using glass fibre filters (Type 13400, Sartorius, Austria; 50 mm diameter, 5–10 mm fibre diameter, #92% porosity), which are Changes in the procedure had to be made when quantitative analyses of evaporative losses of hexadecane were performed.usually used for sampling particulate matter. Filters used in industrial applications have the same range of fibre diameter The PID method was found not to be suitable, because the saturation concentration is very low, so that calibration of the and porosity. High performance fibre filters implemented in industrial device was not possible. For this reason the remaining hexadecane on the test piece after diVerent exposure times to the systems are normally operated at air velocities of 0.04–0.1 m s-1.For measurement purposes liquid aerosols are sampled air flow (120, 240, 360, 480, 720 min) was measured. The test filter was analysed by eluting the organic compounds with with filtration velocities from 0.06 (personal sampling according to BIA 31106) to 0.5 m s-1 (isokinetic sampling using 1,1,2-trichlorotrifluoroethane (quality for IR spectroscopy, Merck, Germany) and IR spectroscopy (PE Spectrum 1000, GSP-PD according to VDI 20668 and BIA 3110, respectively).The experiments covered the range of filtration velocities Perkin-Elmer, Canada). With this measurement method the dynamic behaviour of concentration values cannot be investi- between 0.04 and 0.2 m s-1, which resulted in flow rates of 3.5–12 l min-1.gated, but average values within intervals are feasible. The recommended limit value for the sum of lubricant aerosols and vapour at the workplace in Austria is 20 mg m-3. Results The vapour pressure of n-dodecane at 25 °C results in a saturation vapour concentration of 1100 mg m-3 while n- A typical time-dependent course of the PID-display and pressure drop is shown in Fig. 3. The course of the PID-display hexadecane reaches saturation at the same temperature when an 18 mg m-3 vapour load occurs. The experiments were represents the current evaporation rate. After a steep rise the vapour concentration shows a high level of saturation, then a carried out under zero vapour loads upstream of the tested filter because the maximum evaporation rates should occur flat phase followed by a rapid decrease towards zero.The recorded pressure drop has a decreasing course and reaches a [Fig. 1(b)]. This can be regarded as a worst case scenario. In contrast to industrial demisters where the masses loaded constant value at the end of the high phase of saturation. The final pressure drop of the filter is equivalent to unloaded, dry at fibre filters might reach the saturation point of the material for the retained liquid, the masses at filters used in sampling test filters.The area below the graph of saturation ratio w in Fig. 3 applications are very low. A total MWF mass of 2.1 mg would be expected at the filter if the workplace concentration of represents the evaporated mass, which is equal to the initial mass at the end of the experiment.The initial mass placed on aerosols were 5 mg m-3 and the total sampling air volume using BIA 3110 were 0.42 m3.6 Because of diYculties in the filter determines the width of the high level phase. By using the saturation ratio w, the temperature-corrected dispersing small masses on the one hand and avoiding draino V eVects at the test filter on the other hand, the range value of the saturated vapour pressure pS and the ideal gas equation, the evaporated mass me of the test liquid can be investigated covered initial masses between 6 and 200 mg (4–135 g m-2).calculated by numerical integration over the air volume VL: Temperatures of ventilated air in industrial filter systems are usually between 20 and 30 (35) °C.Sampling of MWF me= PVL r dVL= PVL w pS M R T dVL (1) emissions normally takes place at ambient temperatures within 20 and 25 °C. The temperatures during the experiments were at ambient values. where me [kg]=evaporated mass; R [J kmol-1 K-1]=univer- The experiments were executed with two pure n-alkanes— sal gas constant; w=saturation ratio; T [K]=temperature; pS dodecane (bp=216 °C; vapour pressure=16 Pa at 25 °C) and [Pa]=saturation vapour pressure; VL [m3]=air volume; M hexadecane (bp=287 °C; vapour pressure=0.21 Pa at [kg kmol-1]=relative molecular mass; and r [kg m-3]= 25 °C)—within the range of boiling-point and vapour pressure vapour concentration downstream.of common MWFs. For example, the vapour pressure of By plotting the results of the calculated evaporated mass me Macron H from Shell is about 0.5 Pa9 and for CUT XU from based on measured data against the air volume penetrated OMV the flash-point is specified to be 180 °C.10 Pure substances were chosen to avoid diYculties in detecting complex and unknown mixtures of real MWFs.The developed procedure for carrying out the experiments was as follows: The dry test filter stored at ambient conditions was weighed.In conditions of calm air, the testing fluid was drizzled rapidly onto the filter by using a simple sprayer. Then the initial mass at the filter was determined by weighing the test-piece again. Immediately afterwards the wetted filter was placed in the filter holder and clean laboratory air was pumped through at a defined flow rate.The time-dependent downstream vapour concentration and the pressure drop at the filter and the ambient conditions were recorded automatically. The completed evaporation of the substance on the test filter was indicated by the PID recording and additionally confirmed by gravimetric methods. The accuracy of the PID was checked by sampling the vapour on AmberliteA XAD-2 adsorbent, solvent extraction Fig. 3 Measuring data: PID instrument and pressure drop record. Test fluid: dodecane; air flow: 9.45 l min-1; initial mass 52.1 mg. and quantitative IR analysis according to the analytical J. Environ. Monit., 1999, 1, 373–377 375through the filter the graph given in Fig. 4 is obtained. The evaporated mass me shows a linear increase over a wide range.Fig. 5 indicates the saturation ratios found within the varied range of initial masses. The parameter air flow is represented by diVerent diameters of the bubbles. The saturation ratio w was calculated by using the current concentration r divided by the concentration of saturation rS: w= r rS = 1 rS me VLKprop. (2) For hexadecane, in situ measurement was not applicable.Therefore, the measurement was executed by quantitative IR spectroscopy. For hexadecane, the saturation ratio is given by eqn. (3): w[%]= r rS 100= ma-Mv Vÿ L t rS 100 (3) Fig. 6 Mass transfer coeYcient bmax depending on air flow. where ma [kg]=initial mass; mv [kg]=remaining mass on filter; and t [s]=test duration. Deviations of the overall evaporated mass, calculated by evaporation is restricted by the provided wetted surface.using PID-measurement and checked by gravimetric determi- Furthermore, errors due to the time lag of the measuring nation, did not exceed ±15%. A comparison performed using system probably occurred. quantitative IR spectroscopy confirmed these results. The experiments confirm that loaded fibre filters used for collecting semi-volatile liquids emit retained aerosols as vapour if they are penetrated by unsaturated vapour air flow.As a Discussion and conclusion result of this research, exhaust air of loaded industrial fibre Fig. 3–5 indicate that high rates of saturation occur if loaded filters has to be regarded as saturated with lubricant vapour. fibre filters are penetrated even with clean laboratory air.Over Hence, vapour emissions of industrial fibre filters can be simply a wide range of initial masses and air flow rates, the saturation estimated by calculating the equilibrium load using the saturatio shows no dependency on these parameters (Fig. 5), ration vapour pressure (SVP), the temperature of the ventilated except for very small initial loadings ( less than 10 mg) which air measured at the site and the mean relative molecular mass.result in a decreasing saturation rate. This indicates that Unfortunately, SVP and mean relative molecular mass are seldom specified in the safety data sheet. If only flash-point or boiling-point is given in the safety data sheet a reference value of saturation pressure for lubricants based on mineral oil fractions can be obtained from monographs in e.g.ref. 11. A rough estimation of lubricant vapour production by a typical industrial fibre filter at a 3000 m3 h-1 ventilated air flow gives a value of approximately 3.2 kg h-1 if the vapour pressure of the lubricant is similar to that of dodecane and approximately 55 g h-1 if the vapour pressure is similar to that of hexadecane.15 For very small masses on the filter, which occurs e.g.in measurement applications based on sampling aerosols on fibre filters, the situation might be more complex. Assuming that saturation downstream of the filter does not occur when deposited masses on the fibre filter are very low (e.g. less than 10 mg on a filter of 50 mm diameter), a characteristic, based on the theory of convective mass transfer, is proposed below Fig. 4 Evaporative loss of dodecane from fiber filter. Data points for predicting evaporative losses. calculated by using eqn. (1). Considering the theory of convective mass transfer,12–14 the actual mass-transfer coeYcient b is defined by eqn. (4). Taking the proportional section of the of drawn lines in Fig. 4, a maximum coeYcient bmax can be determined [eqn. (5)]: b= Vÿ L S rS dme dVL (4) bmax= Vÿ L S rS Dme DVLKprop.(5) where b[m s-1]=mass transfer coeYcient; bMAX[m s-1]= maximum mass transfer coeYcient; rs[kg m-3]=saturation concentration; Vÿ L[m3 s-1]=volume flow; and S[m2]=surface of fibres. For hexadecane, where in situ values for evaporative losses were not achievable, data acquired by quantitative IR spec- Fig. 5 Saturation ratio, w, depending on initial mass ma and air flow. Bubble diameters represent air flow. troscopy are used to calculate a coeYcient for mass transfer 376 J. Environ. Monit., 1999, 1, 373–377[eqn. (6)]: References 1 HVBG, Berufsgenossenschaftliches Institut f. Arbeitssicherheit, bmax= ma-mv t S rS (6) BIA Report 7/96, Ku�hlschmierstoVe, St. Augustin, 1996. 2 T. Mang, Die Schmierung in der Metallbearbeitung, Vogl, For the surface S in eqns. (4)–(6), the surface of fibres Wu� rzburg, 1983. calculated by eqn. (7) was used. S was regarded as a material 3 US Department of Health and Human Services (NIOSH), constant and therefore uncertainties in the amount of the Occupational Exposure to Metalworking Fluids—Criteria for a Recommended Standard, NIOSH, Cincinnatti, OH, Publication actual wetted surface have to be modelled by the mass transfer No. 98–102, 1998. coeYcient b. 4 AUVA Merkblatt M 368–1097, Ku�hlschmierstoVe, Unfallverhu �tungsdienst der Allgemeinen Unfallversicherungsanstalt, S=SV (1-e) D2 p 4 B with SV= Sfibre Vfibre = 4 d (7) Vienna, 1997. 5 D. Leith, F. A. Leith and M. G. Boundy, Am. Ind. Hyg. Assoc. J., 1996, 57, 1137.where SV [m-1]=specific surface; e=porosity; D [m]=eVective 6 HVBG, Messung von GefahrstoVen—BIA Arbeitsmappe, Erich diameter of filter piece (clearance of filter holder); B [m]= Schmidt Verlag, Bielefeld, 12. Lfg. IV/94, 1994. thickness of filter; Sfibre [m2]=surface of one fibre; Vfibre [m3]= 7 D. Breuer, unpublished work. volume of one fibre; and d [m]=diameter of one fibre. 8 Verein Deutscher Ingenieure, VDI-Richtlinien 2066 Blatt 2, In Fig. 6, bmax for all the experimental results (near and Du� sseldorf, 1993. below saturation) was calculated. In addition, the theoretical 9 Shell Austria, Sicherheitsdatenblatt gema�ß 93/112/EG zu Material value for saturation is plotted. Fig. 6 shows that most of the 56930, 1996. 10 OMV, Sicherheitsdatenblatt Pd. Nr. 177.615, 1996. calculated values for bmax are near the theoretical maximum 11 Arbeitsmappe fu�r Mineralo�lingenieure, Deutscher Ingenieur value. The points at an air flow of 9.5 l min-1, which are Verlag, Du� sseldorf. significantly below the saturation value, were obtained from 12 H. D. Gaehr and K. Stephan, Wa�rme- und StoVu� bertragung, measurements where the initial mass was at the lower end of Springer-Verlag, Berlin, 1994, p. 340. the range. Further investigations are needed to ascertain 13 H. Brauer, StoVaustausch Einschließlich Chemischer Reaktionen, whether evasurement applications with Sauerla�nder, Aarau, 1971, pp. 110 and 333. low loadings (<5 mg) are significant and if bmax is a proper 14 G. Hanel, Einfu�hrung in die Konvektive Wa� rme- und StoVu� bertragung, Verlag Technik, Berlin, p. 148. characteristic in these cases. 15 B. Riss, Diplomarbeit, TU Wien, 1999. Further investigation is necessary to determine the dependency of evaporation on fibre diameter, ingoing droplet size distribution and the change of composition during evaporation of a complex cooling lubricant. Paper 9/02767J J. Environ. Monit., 1999, 1, 373&ndash