Quality improvement of β-plasters

The binding agent calcium sulfate, consisting of the phases hemihydrate = CaSO₄ · ½ H₂O, anhydrite III and II = CaSO₄, pure or in various mixtures, is produced by thermal dehydration from raw gypsum, e.g. natural gypsum, gypsum from flue gas desulphurisation (FGD) or phosphogypsum = dihydrate CaSO₄ · 2H₂O including impurities. The essential constituent is generally the hemihydrate. This and the anhydrite derived from it occur in an α- or β ‑modification depending on the selected manufacturing conditions. The two hemihydrates have the same crystalline structure but different morphologies and thus also differ in their application properties. α-hemihydrate is produced in a “wet” technical process with high partial pressures of the water steam and possibly with a high total pressure at temperatures in the range of approx. (80-150)°C, e.g. in autoclaves, while β-hemihydrate is the product of a “dry” atmospheric process with a small proportion of water steam in the gas phase and operating temperatures in the range of approx. (120-180)°C. The α-hemihydrate consists of compact, well-structured and relatively large primary crystals, whereas the β -form comprises flaky, jagged, scaly secondary crystals, which are themselves made up of fine individual crystals. High-quality calcium sulfate binding agents, e.g. moulding plaster, are generally manufactured from α-hemihydrate, while mass products, e.g. sandwich-type gypsum plaster board and gypsum wallboard, are generally manufactured from β-hemihydrate. In the following, possibilities are presented for subsequently improving the application properties of a conventionally produced β -plaster.

For reasons of cost effectiveness, the industrial process for dehydration of raw gypsum for the production of plaster of Paris is performed in calciner systems with relatively high burning temperatures, i.e. with large temperature gradients between the heat transfer medium and the charge material and a low partial pressure of the water vapour in the gas phase. This results in extremely fast heating-up and dehydration. For instance, an FGD-dihydrate particle of D_P = 100 μm in a T_F = 160 °C hot gas stream could heat up and transform into b-hemihydrate in approx. ∆t @ 90 ms if the vapour is evacuated immediately. The applicable equation is: ∆t ~ D_P². As the volume of the generated vapour is higher than that of the particles by several powers of ten, a steam overpressure develops in the particles and causes fissures and pores and/or fragmentation. The result is a significantly enlarged, strongly fissured free surface. Like in a grinding process, the stressed areas of the surface are energetically activated.

If the feed material has a broad granulometric distribution, multiphase plasters are produced: the fine dihydrate particles continue to heat up after their transformation to hemihydrate and form anhydrite III, then possibly anhydrite II. In the coarser feed particles the heating processes are delayed/shifted further and further in the direction of the reactor outlet and above a specific particle size they no longer take place. In the case of coarse-grained or lumpy raw gypsum, additional radial phase profiles can form due to the internal temperature gradients.

The processes described are particularly likely to occur in flash calciners and have a significant effect on the resultant product properties and handling characteristics.

As an example, Table 1 compares properties of b-plaster of Paris with those of a-hemihydrate plaster, which is of higher quality but involves a more complex manufacturing process and is therefore more cost-intensive. The shown differences are consequences of the different crystal and surface morphologies of the two types of plaster that result from the production process. Figure 1 illustrates this by means of corresponding CP measurement results. It shows the dependency of the water/plaster ratio WGW, measured according to the strewable quantity process, on the specific surfaces a_BET of the produced/treated plasters determined by the BET method. The curve “FGD plaster I” relates to a FGD-gypsum processed in a flash calciner and subsequently post-processed in different ways in the CP homogenizer described below. The right-hand end point of the curve represents the non post-processed plaster of Paris. As the BET surface a_BET decreases, the WGW also decreases linearly through the measuring range. Whether the homogenizer is operated in continuous or batch mode is of no significance. The curve “FGD plaster II” also shows a linear dependency of the WGW(a_BET). This curve relates to a different FGD raw gypsum that was calcined and post-processed under systematically varied operating conditions in an indirectly heated rotary drum reactor working in batch mode. Due to the different operating parameters, the measured values inevitably show a larger scatter around the best fitting line.

Other properties of the binder materials produced also correlate with the water/plaster ratio WGW, and thus with the surface structure of a plaster of Paris that is characterized by the BET surface a_BET. Examples: the compressive strength R_C and flexural tensile strength P_F increase as the WGW decreases, the applicable equation being (R_C, P_F) ~ 1/WGW², the commencement of setting t_min and end of setting t_max change in line with the WGW etc.

For the rehydration of hemihydrate HH to dihydrate DH 1½ mols of water are necessary. The corresponding minimum water/plaster value referred to the mass of the hemi­hydrate and calculated with the combined water contents w_DH = 0.2092 kg_W/kg_DH, w_HH = 0.0621 kg_W/kg_HH is:

(M_W ) = WGW_min = w_DH – w_HH = 0.1860

M_{HH min} 1 – w_DH⇥(1)

The production of workable plasters demands significantly higher WGWs. The surplus water

∆M_W = WGW – 0.1860

M_HH⇥(2)

in each case has to be dried and if artificial drying is required, as is the case in the manufacturing of sandwich type plaster board, this requires an energy input proportional to the quantity of excess water ∆M_W. On the left-hand ordinate, Figure 2 shows the relative water excess (∆M_W/M_HH) dependent on the WGW while the right-hand ordinate shows the saving in drying ­energy at a lower WGW, referred to the reference value WGW_ref = 0.75, i.e.:

∆E ~ ∆M_W(WGW_ref) – ∆M_W(WGW) = 1 – WGW – 0.1860

∆M_W(WGW_ref)⇥WGW_ref – 0.1860

⇥(3)

Example: if the water/plaster value is reduced from WGW_ref = 0.75 to WGW = 0.60, the energy requirement for drying decreases by ∆E @ 27 %. In the case of a plaster containing soluble anhydrite the above calculation approach has to be suitably modified. However, the energy saving potential remains unaffected.

The explanations provided so far, as well as the comparison of a- and b-gypsums in Table 1, demonstrate that it is good policy to improve the product properties and handling characteristics of b-plasters so that they approach those of an a-plaster. This can in principle be achieved by modifying the manufacturing process of b-plasters or by post-processing the conventionally produced b-plaster Suitable post-processing reactors could be retrofitted to already existing plants. Such reactors should also be able to eliminate or reduce the familiar dependency of plaster of Paris properties on the storage time ( aging), i.e. to produce products with long-term stability.

As shown above, the properties of the gypsum binding agent are significantly determined by its surface structure, characterized here by a_BET. The surface structure is influenced i.a. by the origin and granulometric distribution of the raw gypsum, the particular calcination process used, its operating temperature and the retention time of the plaster in the reactor. Further significant factors for the quality of the resultant product and for the WGW value that is necessary for its further processing are: the phase composition, the particle size and granulometric distribution, the particle shape and the degree of secondary particle disintegration of the plaster when mixed with water. Secondary particle decomposition takes place before noticeable commencement of the actual solution and hydration process and leads to a rise in WGW value due to the increase in freely accessible particle surfaces.

There are essentially two ways of influencing the process in a post-processing reactor:

– artificial aging of the plaster,

– processing under high operating pressure.

Artificial aging: The term aging means the stabilization and/or improvement of the processing properties of the calcium sulfate binding agent through the influence of moisture. If a plaster is spread out over a large area and subjected to a gas atmosphere containing water vapour, e.g. ambient air, the initial reaction is decomposition of AIII to HH. If the amount of available moisture is high enough ( high relative gas moisture content and/or long contact period), the formation of DH subsequently occurs. The prerequisite for this is the existence of liquid water, which is precipitated in the pores and cracks of the particles and in the interstices between the particles as a consequence of capillary condensation and exists as adsorption layer at the free surfaces of the particles. The crystallization of DH takes place primarily at the areas of the particles that are activated by the burning process and leads to a partial filling and closure/covering of cracks and pores. It thus reduces the BET surface. An associated reduction in particle activation also occurs. The WGW decreases, as does the particle decomposition in the mixing water. However, no acceleration of the hydration process has been observed.

The above-described “natural” aging of plaster is not a practicable process in industrial systems as it takes a period of time between several hours and several days before the moisture in the plaster reaches a state of equilibrium. Moreover, the operating conditions vary due to fluctuations in the water vapour content of the ambient air. Particularly if the relative humidity is j ≥ 0.85, considerable absorption of moisture takes place, sometimes over a period of weeks. As the interior of thicker/denser layers of binding agent is less accessible for the humidity than the outer surface, this leads to differing product qualities. In the case of natural aging the operating pressure and temperature cannot be influenced.

However, by causing “artificial” aging, i.e. by supplying a defined quantity of water vapour or possibly water under controlled operating conditions to plaster in a suitably designed post-processing reactor, it is possible to accelerate the aging process and eliminate the described difficulties.

Increased operating pressure: The quantity of water adhering adsorptively to a surface, in this case the accessible surface of plaster of Paris particles, rises with increasing partial pressure of the water vapour p_WD in the surrounding gas atmosphere. This results in a high covering factor, in multilayer adsorption and in increased capillary condensation, i.a. due to modified pore geometries. An increase in the operating pressure p_R of the post-processing reactor while the moisture content of the gas remains unchanged k = M_WD/M_G,tr and the reactor temperature T_R remains constant results in a proportional increase in the partial pressure of the water vapour p_WD and also of the relative gas humidity j. Practical experience has shown that under these operating conditions a further reduction in the BET surface a_BET occurs together with a simultaneous acceleration of the artificial aging process, partly because of the larger availability of water. The measuring points at the left-hand end of the curve “FGD plaster II” in Figure 1 originate from post-processing trials performed with a reactor pressure of p_R (1.75 bar(abs)).

A further possible influencing variable is the post-processing temperature. This has to be harmonized with the upstream calcining process. This point will not be explicitly discussed here.

Supplementary information regarding the above-described relationships is contained in [1-4].

Claudius Peters (CP) operates an industrial-scale gypsum calcining system in its test plant facility. This system has a raw gypsum throughput of M˙_RG ≤ 500 kg/h and can be operated in combination with a calcining mill ( vertical ball mill) or with a flash calciner system. The produced plaster is collected in a fabric filter and then either fed directly to a storage silo or alternatively to the post-processing reactor. The design and functioning of this reactor, which is called as a “Homo­genizer”, are discussed in detail in the following. Previous investigations were carried out with a continuously operating homogenizer under atmospheric pressure [5-9]. At the beginning of 2011 this was replaced with an identical-type, but optimized design working in continuous operation that can achieve reactor overpressures of up to p_R = 3.0 bar overpressure [10]. Corresponding trials are currently taking place. The homogenizer trials are accompanied by tests using a laboratory homogenizer with a volume of V @ 20 l and an electrically-heated rotary drum calciner [11, 12]. The latter equipment can also be operated under higher operating pressures if required.

Figure 3 is a schematic representation of the homogenizer design that is capable of high pressure operation. The plaster to be processed is fed quasi-continuously to the reactor via a lock hopper system and extracted in an identical manner. During operation at atmospheric pressure these components are either removed or set to continuous operation. By means of a controlled pressure maintaining valve the pressure in the upper reactor compartment is adjusted to the selected operating value p_R and held at that setting. The actual homogenizer is designed as a fluidized bed reactor with forced internal circulation. This is performed by a central circulation pipe with its own gas supply. The maximum permissible filling level of the fluidized (expanded) bulk material in the outer reactor compartment is approx. 300 mm below the upper outlet of the central pipe. To allow the setting of different mean bulk material retention times

t{ = ∆M_G

M˙_G⇥(4)

with: ∆M_G – current plaster mass in the homogenizer,
M˙_G – plaster mass flow through the homogenizer,

the height of the fluidized bed (~ plaster mass ∆M_G in the reactor) can be set to lower values by means of a vertically adjustable filling level gauge. The bulk material is discharged from the homogenizer via an outlet pipe, whose inlet height over the distributer bottom can also be varied.

Usual b-plasters of Paris can be classified in the Geldart diagram [13] as group A to A/C bulk materials, i.e. they are relatively easy to fluidize and initially expand homogeneously above the minimum fluidization velocity u_F,L through a limited gas velocity range before bubbles start to form. Although the bubbles formed are limited in size ( they disintegrate into smaller bubbles once they have exceeded a critical diameter), they do assist mixing of the bulk material in the homogenizer. In the outer compartment of the reactor the selected fluidizing gas velocities are therefore above the respective bubble formation velocity u_F,B. This velocity can be calculated [14]. Typical empty pipe gas velocities for plasters are in the range of u_F,R @ 2 m/min and are independent of the reactor pressure p_R. Gas velocities of u_F,C ≥ (2.5 · u_F,R) are selected for the circulation through the central pipe.

The water vapour/water required for the post-processing can be supplied in different ways:

1. A partial flow of the exhaust gas containing water vapour from the upstream calcination process or another source can be used as fluidizing and circulation gas. Advantage: uniform distribution of the water vapour over the cross-section of the fluidized bed; no additional energy requirement for water evaporation. Disadvantage: the amount of water vapour in the gas is often not sufficient for optimum post-processing of the plaster.

2. Supplying of water vapour via a special steam nozzle installed in the central circulation pipe, injecting the steam in the upward direction of conveyance, see Fig. 3. Advantage: the amount of steam can be flexibly adjusted to requirements. Disadvantage: the water vapour has to be separately generated.

Supply variants 1. and 2. can be advantageously combined.

3. As 2., but supplying liquid water instead of water vapour via a suitable spraying/atomiser nozzle. Advantage: no generation of water vapour required. Disadvantage: distribution of the water in the bulk material is extremely problematic, gypsum caking of the riser pipe wall etc.

Supply variants 1. and 2. were successfully applied in pilot plant tests. Variant 1. has already been implemented in two industrial systems while variant 3. is being tested.

If implementing supply variant 2., it has to be ensured that every plaster particle passes the steam nozzle in the circulation pipe, i.e. is in contact with steam, at least once, but naturally better several times, during its retention time in the reactor. The same applies to variant 3. In this connection, the circulation number

U = M˙_G,C

M˙_G⇥(5)

is defined. This specifies the circulation mass flow M˙_G,C that is required for the circulation of the reactor contents “U” number of times during the bulk material retention time t{. The value “U” is specified for the reactor design. On the other hand, M˙_G,C can be explicitly represented as a function of the peripheral conditions of the plaster conveyance. The available pressure difference ∆p_C for circulation through the central pipe is determined by the gas pressure drop ∆p_R of the fluidized bed in the reactor’s outer compartment, as both compartments are connected at the bottom and at the top; therefore: ∆p_C = ∆p_R = H_R · r_b,R · g. If only the dominating pressure loss due to lifting the solids is applied as the conveying pressure loss ∆p_C, the applicable equation is:

M˙_G,C = C · H_R · r_b,R · M˙_F,C

H_C r_F,C⇥(6)

where: g – (= 9.81 m/s²), acceleration due to gravity,

C – velocity ratio solids/gas in the circulation pipe,

H_R, H_C – height of fluidized bed, height of circulation pipe, measured above the distributor bottom,

r_b,R, r_F,C – mean (fluidized) density of the fluidized bed, mean gas density in the circulation pipe,

M˙_F,C – conveying gas mass flow through the circulation pipe.

The precise solution also includes, inter alia, the acceleration and friction losses of the bulk material and the conveying gas.

The conveying gas flow M˙_F,C needed for the required plaster circulation mass flow M˙_G,C can be calculated from equation (6) or the precise solution and the diameter D_C of the circulation pipe can be calculated by means of the selected gas velocity u_F,C. The number of circulations U is defined as U ≥ 2. Up to now, it has not been observed that the supply of water vapour causes coatings to form inside the circulation pipe. This is prevented by the 2-phase mixture flowing through the pipe.

Characteristic values and their dependency on the respective operating condition, e.g. on r_b,F(u_F,R), are required for designing the reactor. The calculation approaches and models proposed in the literature for this purpose are only of limited precision and are therefore checked by laboratory experiments for design purposes. Overall, the operating behaviour of the reactor permits pre-calculation and safe scale-up. A newly developed design program enables the performance of detailed analyses [15].

The homogenizer can be equipped with so-called agitators to mechanically assist the formation of a fluidized bed if the plaster is cohesive and very difficult to fluidize. This can be necessary, for example, in the case of products from the left-hand section of Geldart group C. Alternatively, a pulsating active/inactive bottom aeration can be used. With some sorts of plaster, the handling properties and product characteristics can be improved by intensive mechanical stressing/activation such as  that rendered by the agitators. This particularly applies to phosphoric gypsums.

The main technical data of the CP test homogenizer are: fluidized bed diameter D_R = 1.0 m; maximum fluidized bed height H_R,max @ 1.3 m; overall height of reactor H_ges @ 2.6 m; inside diameter of circulation pipe D_C = 212.7 mm; permissible reactor overpressure p_R = 3.5 bar overpressure; plaster throughput M˙_G ≤ 500 kg/h; reactor completely thermally insulated with electrical heating underneath the insulation; fluidization gas and circulation gas heated up by electrical heating coil in the supply pipe; electrically heated water vapour generator; including pressure maintenance, inlet and discharge systems; complete set of instruments with control system and integration into the higher-level calciner system.

Up to now, homogenizers installed in industrial plants have been equipped with fluidized bed diameters of D_R = 2.2 m and 3.6 m.

3.2 Results

As an example, measurement results are presented that were obtained with the FGD plaster I shown in Figure 1 which was produced in a flash calciner at approx. T @ 160°C. For these tests the pressure p_R in the upper reactor compartment was identical with the ambient pressure. Water vapour was supplied at a controlled rate to the circulation pipe via the steam nozzle. Figure 4 shows the dependency of the water/plaster ratio WGW ( left-hand ordinate) on the added specific quantity of water vapour

X_WD = ∆M_WD

∆M_G⇥(7)

Starting at WGW(X_WD = 0) @ 0.61, the WGW decreases monotonously in line with the rising X_WD and above the value X_WD @ 60 g_WD/kg_G it settles within the limits of measuring accuracy at a constant value of WGW @ 0.55. This corresponds to a reduction of approx. 10 % in the water/plaster ratio.

Figure 4 also shows the change in the combined water content RKW ( right-hand ordinate) caused by adding water vapour X_WD. The plaster from the upstream calciner system has RKW(X_WD = 0) @ 5.20 M.-%. With the measured raw gypsum purity of 97.34 M.-% and assuming a pure hemihydrate phase a theoretical RKW(HH) = 6.04 M.-% is obtained. The curve RKW(X_WD) in Figure 4 makes clear that first the obviously present anhydrite III is converted to hemihydrate and that dihydrate is subsequently formed, as the combined water content rises past the value RKW(HH) = 6.04 M.-%. to approx. RKW @ 6.50 M.-%. The measured BET surfaces in Figure 1, curve FGD plaster I, confirm that the new formation of dihydrate caused by the (adequate) water vapour treatment of the plaster in the homogenizer leads to a “healing” of the cracked/fissured surface structure that had been caused by the upstream calcination process. This primarily takes place on the activated surface regions of the particles. Measured granulometric distributions of plaster samples taken upstream and downstream of the reactor confirm this by showing no differences or changes resulting from the post-processing.

Other effects of the water vapour treatment/artificial aging on the properties of the post-processed plaster are shown in Figure 5. It depicts the setting times t_min, the initial setting time measured by the knife-cut method, and t_max, the final setting time measured by the Vicat method, as a function of X_WD. Compared to untreated plaster, t_min and t_max are lengthened, in part significantly, by the addition of water vapour. In addition, the working range (t_max - t_min) broadens in line with increasing X_WD.

The compressive strength R_C and flexural tensile strength P_F that were measured in previous tests follow the above-described relationship (R_C, P_F) ~ 1/WGW² with good agreement, i.e. both increase in line with decreasing WGW.

Figure 6 shows the first measurement results of plaster postprocessing at increased reactor pressure as a function of the WGW(a_BET) ( left-hand ordinate). The FGD plaster III, was post-processed at p_R = 1.2 bar overpressure in the upper reactor compartment. Further boundary conditions were: plaster supplied by an upstream flash calciner, continuous system ­operation, and constant water vapour supply at X_WD @ 62 g_WD/kg_G. The mean plaster retention time t{ was varied. In Figure 6 the water/plaster ratio decreases in line with rising t{, i.e. the healing process takes a certain time, which has to be made available. It is obvious that this process takes place faster at increased reactor pressures than in the homogenizer tests at atmospheric pressure. On the right-hand ordinate of Figure 6 the plaster particle density r_P measured in the gas pygnometer and modified by vapour treatment for the same test series are plotted as a function of a_BET. After an increase due to the transformation from AIII to HH, r_P decreases in line with the declining specific surface a_BET. This confirms that the new formation of dehydrate causes the open cracks and pores of the plaster particles to become covered over. With approximately constant particle mass, this covering process increases their apparent volume, i.e. the density r_P decreases.

If water vapour specially generated for the post-processing is used alone or in combination with an exhaust gas containing water vapour, then a simplified heat balance shows that the work for this vapour generation only produces an energetic benefit, for instance through reduced energy requirement for drying the plasterboard, if the condition

X^*_WD < 1

∆WGW⇥(8)

with: X^*_WD = quantity of water vapour specially generated for the steam treatment,

∆WGW = WGW (before homogenizer) – WGW ­(after homogenizer), the WGW reduction achieved through the post-processing,

is observed. This criterion is based solely on energetic considerations; advantages resulting from improved product quality are not considered here. The use of exhaust gas containing water vapour for the steam treatment reduces or in some cases eliminates X^*_WD.

Up to now, the CP homogenizer has been used with success for post-processing a variety of FGD, phosphoric and natural gypsums. The very considerable operating experience thereby gained is a decisive factor in the design of production plants. The employed hardware, particularly the lock hopper systems for continuous gypsum feeding and discharge when the reactor is operating under increased pressure, has proved effective.