1/7 HICOSYS FOR PROCESS CONTROL OF SPRAY DRYERS CAPACITY INCREASE AND FOULING REDUCTION BY MEASURING HUMIDITY OF DRYING AIR? In spray drying processes capacity of the dryer is partly determined by the humidity of the drying air. The more water the inlet air contains, the lower the remaining capacity for water uptake of this air is. By measuring humidity of the drying air the process can be controlled in a way to maximize water uptake of each kilogram of air. At the same time powder production capacity is maximized and energy efficiency is increased. A main issue of spray drying is fouling of the dryer due to stickiness of the product, especially for products with a high content of carbohydrates. Fouling results in shorter running time between cleaning procedures and in the worst case causes blocking of the dryer by lumps of powder. Improved process control by humidity measurements can minimize fouling, extend the time between CIP s and avoid blocking of the dryer. This will again result in capacity increase and also safe processing. HOW TO MEASURE HUMIDITY? The sticky point of the product as a function of the temperature can be determined at lab-scale, representing commercial operating conditions. The sticky temperature can be plotted against the relative (RH) or absolute humidity (X) of the surrounding air. An example of such plots, both based on the same powder, is shown in Figure 1. 120 Relative humidity versus Stickiness Tstick ( C) 80 40 not sticky sticky 0 0 10 20 30 40 50 RH (%) 120 Absolute humidity versus Stickiness Tstick ( C) 80 not sticky sticky 40 0 0 8 16 24 32 40 X (g/kg) Figure 1. Sticky temperature versus relative humidity (a) and absolute humidity (b).
2/7 CFD (Computational Fluid Dynamics) calculations pointed out that in a spray dryer there is a jet of hot and dry air just below the air inlet and in the remainder of the dryer the air conditions are very close to the outlet air conditions. The spray drying rate is limited by water diffusion within the particle. The outer shell of the particle dries very fast and reaches the equilibrium moisture content with the surrounding air after a short drying time, while the inner of the particle is still wet. The condition of the outer shell of the particle in equilibrium with the outlet air will determine if the product is sticky or not sticky. Therefore it is important to know the temperature and humidity of the outlet air. The temperature can be measured easily, but measuring the humidity is more problematic. Capacitive humidity sensors do not like temperatures above 70 C and have an inaccuracy of ±1-2% RH, which is too high for a good measurement of the outlet air with a RH around 10%. Also indirect measurements (e.g. Mollier diagram or heat and mass balances) are not accurate enough. The Mollier diagram will fail because the spray dryer is not fully adiabatic, because there is an unknown heat loss to the environment. Heat and mass balances will fail because there is no accurate measurement of the flow rate of the main air. Another complicating factor is that a lot of two stage dryers have an open connection between the drying chamber and the external fluid bed. If there is an unknown air flow from the external bed to the drying chamber, all indirect measurements get even less reliable. The conclusion is that a direct and accurate measurement of the humidity of the outlet air is highly preferred. Besides the detection of CO for pre-warning for fire and explosion, the HICOSYS provides an accurate measurement of the absolute humidity of air (± 0,3 g/kg). STICKINESS AND CAPACITY To keep the powder in the non-sticky state and at the target moisture content, the right in- and outlet air temperatures of the spray dryer have to be adjusted. The optimal temperatures also depend on the absolute humidity of the inlet air. A higher humidity will need a lower inlet air temperature, resulting in a lower capacity of the dryer. For sticky products an increment of the absolute humidity of the inlet air with 1 g/kg will result in a capacity decrease of 3%. As shown in Figure 2 there is a seasonal variation in the absolute humidity of atmospheric air, but there are also short term variations. If the variations are not taken into account and constant process settings are applied, the settings should be in such a way that also during peaks in humidity the spray dryer acts in the non-sticky state. As a consequence, the capacity is far below optimal during periods of a low absolute humidity of atmospheric air. For example, if the absolute humidity is 5 g/kg below the peak value, there is about 15% loss of capacity.
3/7 short term variations seasonal variations Figure 2. Typical annual course of absolute humidity of atmospheric air (Source: KNMI, NL) The nose shaped stickiness-curve as shown in Figure 1b (X vs T) is very suitable for process control: the outlet air conditions (T, X) have to be in the non-sticky region and the absolute humidity (X) should be as high as possible, because in that case the highest capacity is reached. So, the highest capacity is reached if the outlet air condition is in the tip of the nose. However, the plot does not indicate the corresponding powder moisture content at that conditions. To find the settings for optimal capacity at the target powder moisture, the sorption properties of the product should be taken into account and a more advanced tool is needed. NIZO food research has developed such a tool called Premia-DrySPEC3. A few screen prints are shown in Figure 5 a-c. This tool can be applied off-line as well as online. Premia is supplied with OPC (Open Platform Communications) client software to read data from process control equipment and connect the data to the DrySPEC3-model. In this way the effect of changes in weather conditions on optimal settings can be made visible for the operator and it becomes possible to always run at optimal capacity without risk for fouling.
4/7 Figure 3a) Screen print of Premia-DrySPEC3. Process -tab : Processing conditions of test 1 on NIZO pilot spray dryer (yellow: on-line measured; green: input fields; blue: calculated fields) Figure 4b) Screen print of Premia-DrySPEC3. Process control -tab: Overview of current and optimal conditions.
5/7 120 100 Absolute humidity versus Stickiness Red = sticky; Green = non-sticky Drying chamber Drying chamber optimal IFB Tstick ( C) 80 60 40 20 0 10 20 30 40 50 60 X (g/kg) Figure 5c) Screen print of Premia-DrySPEC3. Plot of stickiness curve, current positions and optimal condition regarding capacity and desired powder moisture. TESTING ON PILOT SPRAY DRYER The HICOSYS is installed and tested on the pilot spray dryer of NIZO food research. The multi stage dryer (NIRO-250) is equipped with an internal fluid bed and an external vibro-fluid bed. Eight sample probes are mounted at locations as shown in Figure 6. With probe 5 the absolute humidity of the outlet air of the dryer is measured. Figure 6. Locations of sample probes on pilot spray dryer at NIZO food research. The DrySPEC-tool was on-line fed with real time data of the pilot spray dryer. Five tests were performed with whole milk powder (see Table 1).
6/7 In test 1, the dryer was started and run until steady state was reached at conditions that were for sure in the non-sticky region. With DrySPEC the advised optimal settings (see Figure 5b) were considered and adjusted for test 2. Under these conditions the feed flow could be increased and no sticky issues were encountered. This test was taken as a reference for the next three tests. To simulate changes in weather conditions, in test 3 and 4 the humidity of the inlet air was increased by injection of a controlled flow of steam. In each test, the dryer temperature conditions as proposed by the DrySPEC-model were set. This resulted in a lower capacity, but no stickiness issues were encountered. In test 5 the reference conditions of test 2 were set. The humidity of the inlet air was increased to 18.9 g/kg, but the settings of the dryer were not adapted. Now, serious stickiness issues arose: in the internal fluid bed lumps of powder were formed and finally the bed got fully blocked (figure 5). The explanation is that the powder reached the internal fluid bed in the sticky state, causing a bad fluidisation of the powder leading to lump formation. Table 1. Tests on pilot spray dryer Test X main air (g/kg) T air in ( C) T air out ( C) X air out (g/kg) Feed flow (kg/h) 1 9.6 188 86.5 37.6 329 2 (ref) 9.5 200 77.2 48.7 415 3 13.9 189 78.1 47.8 368 4 18.5 177.5 78.3 47.8 340 5 18.9 200 77.5 55.5 395 Figure 7. View on top of blocked IFB. IFB completely filled with powder and only at the outlet rotary valve powder some powder was removed.
7/7 These tests demonstrate the benefit of adapting the dryer conditions when inlet air conditions change. When running all the time the same dryer conditions, the conditions of test 4 should be applied to guarantee a safe production without stickiness issues under all weather conditions. When monitoring the air humidity and adjusting the dryer settings, the dryer capacity of whole milk powder production can be increased up to 415 kg/h which is 22% higher than under the conditions of test 4 (340 kg/h). Infant formula powders are more sticky than whole milk powder and experience has learned that outlet air humidity s not higher than about 30 g/kg can be applied. For such products the differences in dryer capacity are even higher. The conclusion is that an accurate measurement of the humidity of the inlet (probe 1) and outlet air (probe 5) is very useful to watch over changing weather conditions. Undesired changes in the absolute humidity of the outlet air will become visible and actions can be taken to avoid stickiness issues. The DrySPEC-tool can be very useful to find the needed adjustments to get back to the optimal state regarding stickiness and powder moisture. Also the benefit of the other probes was examined. Possibly probe 8 could give useful information about the drying behaviour in the internal fluid bed, in case it measures the air out of the internal fluid bed. The practice showed that in case the probe was very close to the powder bed it was sensitive for fouling and in case it was located at a safe distance from the powder bed the reading was affected by the humidity of the main air. So mounting of this probe is not recommended. The air measured by probes 2, 3 and 7 are originated from the same fan room. The readings are almost exactly the same, so in such a case 1 probe is sufficient. Probes 4 and 6 are useful for detecting smouldering parts (CO-production) in the external fluid bed and the second cyclone and are not needed for process control.