Some Data on Heat Transfer in Multiple Effect Evaporators

Heat transfer data from measurements covering 15 Roberts evaporator vessels at 3 mills (82 data sets total) is presented. Heat transfer coefficient (HTC) is plotted against brix, temperature, temperature difference ( At) and viscosity respectively. It is shown that the influnce of At on HTC cannot be deduced from this type of data. The pronounced difference in operating conditions between last effect and all earlier effects, due to viscosity levels which are orders of magnitude higher in the las, effect, is highlighted. HTC values from second to penultimate effect nearly all fell into the range of 1,8 to 3,5 kW m-z OC-I, with no pronounced dependence on effect number. It is shown that very high last vapour vacuum levels are detrimental to evaporation capacity and optimum last vapour saturation temperature is in the range 55 to 60°C. A heating surface distribution wherein last effect is double the size of intermediate effects would provide 6% more evaporation than the conventional arrangement of all tail vessels having equal area. HTC values are presented for evaporator sets where this has been achieved by paralleling existing vessels on vapour. It is ooncluded that for series juice feed, the advantage,of lower brix in the first vessel of the pair is offset by the absence of flash in the second vessel. Introduction In recent vears a number of surveys have been undertaken to determini heat transfer coefficiehts (HTC) of the individual vessels in multiple effect evaporators in Huletts mills. The main purpose of this work has been to identify and quantify instances of sub-optimum performance. In some cases severe steam-side tube fouling conditions were found which were corrected by chemical cleaning (Lewis et ale). In the course of this work a fairly substantial body of heat transfer data has been accumulated. The purpose of this paper is to discuss certain indications that can be derived from these results, in respect of both operating conditions and evaporator configuration. Some cautionary remarks on the dangers associated with interpretation are also included. Measurement and Evaluation Procedures The procedure used for determination of heat transfer data varied somewhat between different investigations, but the foll~owing basic approach was used to derive all the results presented here. The main data required for calculation of evaporator overall heat transfer ooefficients are heating surface, evaporation rate and temperatures on both shell and tube side of the calandria. The most practical procedure for deriving evaporation rates is to determine flow rate of solids or brix through the train and measure juice concentration (as brix %) at all terminal and intermediate points. This was the approach adopted here. Brix rates were derived from mixed juice mass and analysis data, adjusted for loss of brix in the filter cake output from the clarification stage. Test run duration was typically two to four hours to enable fluctuations in flow rates and pressure conditions to be averaged out. Juices and syrup were sampled continuously, through cooled copper coils to eliminate sample flash or evaporation, into large glass containers which were under partial vacuum where necessary. Temperatures can be measured either directly or indirectly via pressure readings. Our experience with direct temperature measurement has not been good. Values obtained were frequently lower than saturation temperatures derived from simultaneous vapour pressure measurements, even though thermometer pockets in juice and condensate outlet lines were located as close to the vessel as possible in order to minimise the effect of flash cooling. Therefore only data derived from measurements of pressure in the calandria and the evaporator body has been included here. Pressure measurements were made with a mercury manometer in most cases, or alternatively with calibrated gauges, and corrected for barometric pressure. Corresponding saturation temperatures were adjusted for boiling point elevation and hydrostatic head on the juice side, but not fmor condensate subcooling on the steam side. All data processing was done using the evaporator calculation program PEST (Hoekstraz). In addition to the factors already mentioned, this program takes into account temperature/enthalpy relationships, juice superheat, vapour bleeding and the return of condensate flash in the precise configuration applicable to each installation. Results and Discussion Data from 19 test runs covering a total of 15 vessels at three mills is shown in the figures discussed here. This data has been screened to exclude results for vessels which are known to suffer from steamside tube fouling, and for cases of two-vessel effects. The latter situation is abnormal if juice feed is in series since there is no flash in the second of the pair, and if juice is fed in parallel then there is a problem with quantification of the feed split and hence determination of evaporation rates. A few more results were rejected because of suspected dilution or contamination of juice samples. Long tube climbing film evaporators are also excluded because of their fundamentally different design. Many of the first effect vessels in the installations studied here are of this type. It so happened that data for most of the remaining first effect vessels had to be,excluded for one or other of the reasons mentioned above. The evaluation has thus been restricted to tail vessels only, that is from second to last effect. This has the further advantage that, with one exception, no vapour bleed is withdrawn from the vessels considered and one potential source of error is thus avoided. Of the five evaporator trains represented, four are quadruple effect and one quintuple effect. The vessels have been divided into three groups, represented by different symbols in the figures, as follows : 0 Second effect A Intermediate effects third effect, plus fourth effect of quin Last effect seedings of The South African Sugar Technologists' ~ssocration June 1981