Use of the CP g-cooler in waste heat
recovery systems

1 Introduction

The Claudius Peters g-cooler is a moving bed reactor for indirect cooling of free-flowing bulk materials. The thermal content of the bulk material is transferred to a coolant flowing through a pipe bundle with specially shaped and arranged pipes. Originally, the g-cooler has been developed for cooling cement clinker down from t_K,in ~= 400°C to t_K,out ≤ 100°C using air as the cooling medium. For this purpose, a short grate cooler/recuperator is installed downstream of the rotary kiln, to cool the clinker down from the kiln outlet temperature of t_K,kiln ~= 1350°C to approx. t_K,in, and to allow almost complete utilization of the high-temperature cooler exhaust air in the kiln system, except for a control reserve. The ambient air fed to the g-cooler as cooling air is discharged from the cooler in a clean, dust-free condition and therefore does not require cleaning. For this reason, there is no need for complex exhaust air filtering equipment. Alternatively, it is possible to utilize a g-cooler for increasing the throughput of a cooler system by installing it downstream of an existing conventional (long) grate cooler. Up to now, 23 g-coolers have been supplied for the applications described above.

Claudius Peters Projects GmbH analyzed the usage of a recuperator/g-cooler combination for increasing the energy efficiency, protecting the environment and conserving resources in cement plants [1]. One of the main points of study was whether the existing process of clinker-heat extraction for use in a downstream power generating plant, involving a conventional waste heat exchanger/waste heat boiler, could also be realized directly in a g-cooler using a suitable heat transfer medium. Figure 1 compares these two variants: design (b) utilizes the thermal content of the clinker directly for heating up the heat transfer medium and thus permits the achievement of higher heat-transfer-medium outlet temperatures, i.e. delivers a higher-value heat flow. Moreover, the amount of required equipment appears to be lower than it is in design (a).

The required independence of the clinker burning and cooling process from malfunctions of the power generating plant rules out the direct application of the medium used for the power generation, e.g. water/steam, as the transfer medium for heat extraction. This prerequisite also demands the installation of an emergency system for evacuating the clinker heat in the event of failure of the power generating plant. In the following, the power generating plant will be regarded as an autonomous system that is connected to the clinker cooling system by means of a separate heat-transfer-medium circuit. Furthermore, the considered method of extracting the clinker heat from the g-cooler will be a heat-transfer-oil circuit. In the considered system, the employed oils remain liquid over the entire working range of t_WT,in to t_WT,out. Their boiling temperature, i.e. the temperature at which boiling commences, is higher than t_WT,out at a pressure of p ~= 1.0 bar(abs.). This means that it is not necessary to work with high process pressures, such as those required in water/steam systems. This simplifies the safety engineering and reduces the necessary capital expenditure and operating costs.

Selection of the heat transfer oil for use in the heat extraction system depends on the expected maximum operating temperature. This is the temperature at a kiln flush at which a short-term surge occurs in the amount of clinker leaving the kiln. This results in insufficiently cooled clinker and temperatures in excess of the normal outlet temperature t_WT,out. Since the reaction of a g-cooler heat-transfer-oil system to this event is different to that of a conventional waste heat exchanger, this operating case has to be analyzed separately.

Aside from the size of g-cooler needed for the special application and the required heat transfer area, other questions to be resolved in addition to the already indicated problems include the following:

The effect of a pipe breakage = oil leak in the g-cooler

The flow guidance of the heat transfer medium in the cooler that is necessary in order to maintain operation in the event of failure of plant sections, e.g. due to a clinker blockage or oil leak

Consequences of variation of process parameters

Monitoring of the cooling process

etc.

The following considerations mainly relate to the process-engineering dimensioning of the g-cooler, but also take account of some of the above-mentioned questions. All the results are represented on the basis of the following reference case: clinker throughput M˙_K = 5000 t/d = 57.87 kg/s; clinker inlet temperature t_K,in = 420°C, clinker outlet temperature t_K,out = 100°C; heat-transfer-medium inlet temperature t_WT,in = 50°C, heat-transfer-medium outlet temperature t_WT,out = variable.

2 Process engineering of the g-cooler

In the following the original cooler design, improvement modifications, the heat transfer from bulk material to pipe wall and the thermal design are presented in summarized form.

2.1 Cooler design

Figure 2 is a schematic representation of the equipment configuration required for g-cooler operation. The clinker that has been broken-up by a roller crusher at the end of the recuperator is transported to the cooler by means of a hot-material bucket apron conveyor, and distributed over the length of the cooler by a drag chain. The drag chain transports the clinker over a bar grate, which limits the maximum size of clinker that is transported to the cooling zone/the pipe bundle to d_K,max ~= 30 mm. The oversize clinker is conveyed to a special discharge device. Conveyance of the clinker away from the cooler outlet is handled by oscillating troughs. The oscillation frequency of these troughs is infinitely variable, and is the factor defining the velocity u_K at which the clinker moves down through the cooler. The cooled clinker falls onto a belt conveyor, which transports it to the storage facility.

The pipe bundles of the cooler consist of lens-shaped pipes with the dimensions shown in Figure 2, detail X. Due to the geometry of the pipes and to their spacing, the bulk material flows through the pipe bundle without the formation of bridges and dead zones. The pipe bundle dimensioning is based on knowledge of the statics and kinematics of bulk materials [2]. The rows of pipes are arranged in a staggered configuration. This prevents the formation of a steady-state temperature profile, which would be detrimental to the heat transfer between the bulk material and the pipe walls, see section 2.2. For reasons of wear, the downward movement of the clinker is limited to a speed of u_K ~= 4.5 cm/min, related to the free cross-section between the cooling tubes. The pipe bundle is of modular design, comprising identical compartments arranged one above the other in parallel columns. One cooling air fan is assigned to each column. The cooling air passes through the L_R = 3.0 m long pipes in cross-counterflow to the clinker. Continuous filling level measurement prevents overfilling. Further details are provided in [3, 4, 5].

Although the lens-shaped pipes employed up to now provide optimum flow characteristics, they are complicated to manufacture and unsuitable for higher operating pressures. Their replacement with commercially available round pipes has the consequence that bulk material accumulates in a stagnant zone against the flow-impact side, while a gas void without material forms on the rear side, see Figure 3. In these stagnant zones, the heat has to be transmitted to the pipe wall by conduction. This results in worse heat transfer than at the pipe walls that have direct contact with the flowing bulk material. On the basis of mechanical and physical properties of bulk materials, the geometry of the stagnant zones and the associated reduction of heat flow can be estimated. The studies therefore also considered the usage of round pipes, here DN 100 ––– (Ø114.3 x 6.3) mm, with suitably adapted geometrical arrangement. At the same time, the pipe length was increased to L_R = 3.6 m. Other structural modifications will be described later.

2.2 Heat transfer from bulk material to pipe wall

As the heat transfer through the pipe wall and to the liquid heat transfer oil flowing through the pipes are well-known standard problems, see e.g. [6], the following passages will deal solely with the main aspects of the more complex heat transfer α_K,tot from non-gas-permeated bulk material to a cooled pipe wall [6-11]. The total heat transfer α_K,tot with the associated temperature difference (t_K – t_K,w) comprises two individual resistances connected in series, see Figure 4: These are the contact resistance (1/α_K,w) of the layer of particles in direct contact with the pipe surface, temperature difference (t_K,p – t_K,w), and the internal thermal resistance (1/α_K,bed), temperature difference (t_K – t_K,p), of the actual bed of clinker. The following applies:

————   1   α_K,tot = ————   1   α_K,w + ————   1   α_K,bed⇥(1)

α_K,w defines the direct heat transfer to the cooling pipe, while α_K,bed describes the subsequent transportation of heat from within the bed of material. The equations required for calculation of α_K,w are provided in annex A. They describe the heat transportation via the contacts of material particles and pipe wall and via the gas interstice between them, which is caused by the particle shape and the surface roughness. α_K,w provides maximum values in the case of very short contact times between the bulk material and the pipe wall. These values represent the upper limit for α_K,tot [9].

For calculation of the subsequent heat transportation through the bed of material, the bed is treated as a continuum with correspondingly defined effective characteristics. Details of this are provided in annex B. The contact time τ_K,w of the clinker bed and the cooled wall required by equation (B1) is calculated as contact time at the individual pipe, as the bed of bulk material can be regarded as thoroughly thermally mixed because of the alternate heat extraction due to the staggering of the individual rows of pipes. As a consequence, no steady-state temperature profile occurs. With the maximum downward-movement speed u_K ~= 4.5 cm/min, the contact time calculated for the standard lens-shaped pipe is τ_K,w ~=  510 s. This decreases to τ_K,w ~=  210 s in the case of the DN 100 pipe.

The total heat transfer coefficient k from the bulk material to the heat transfer oil is derived from the equation

—  1 k = ————   1   α_K,tot + ————   1   α_w   + ————   1   α_WT⇥(2)

with:

α_w= λ_w /∆s_w = heat transfer through the pipe wall,

λ_w = thermal conductivity of the pipe wall material,

∆s_w = pipe wall thickness.

In Figure 5, the different heat transfer coefficients calculated for a given operating case are plotted from the bottom to the top along the height/pipe levels of a g-cooler [12]. The depicted operating case is the reference case with a heat transfer medium outlet temperature of t_WT,out = 300 °C, a heat transfer oil selected to suit the given temperature range and pipes of type DN 100. It is known that the total heat transfer coefficient k is always smaller than the smallest of the individual heat transfer coefficients in equation (2). Here, the heat transfer at the bulk-material side described by α_K,tot is lower by powers of ten than (α_WT, α_w), and is thus the limiting mechanism for the transportation of heat. The k-value is almost identical with α_K,tot. For this reason, improvements in the heat transfer have to take place on the bulk-material side; in the case of a liquid heat transfer medium, the transfer-medium side has no influence. Comparable results were found in all the conducted calculations.

2.3 Calculation model

Among the information required for designing a cooler are the clinker data (M˙_K, t_K,in, t_K,out), the heat transfer medium data (t_WT,in, t_WT,out or M˙_WT) and the associated material properties. From the total thermal balance

Q˙ _WT = M˙_K · c_p,K|t_K,in   t_K,out · (t_K,in – t_K,out)

= M˙_WT · c_p,WT|t_WT,out   t_WT,in · (t_WT,out – t_WT,in)⇥(3)

with:

c_p|t_out   t_in  = mean specific heat between t_in and t_out,

the heat flow Q˙ _tot to be discharged, the heat transfer medium flow M˙_WT required for this purpose or, if M˙_WT is given, the resultant outlet temperature t_WT,out can be calculated. Because of the changing heat-transfer conditions/temperatures along the height of the cooler, each pipe level is separately considered, i.e. each with separate mean values. The following equation then applies:

Q˙ _tot = ∑   ^N   _i=1 ∆Q˙ _i = ∑   ^N   _i=1 (k · A · ∆t_αν)_i⇥(4)

with:

i, N = i^th pipe level, total number of pipe levels,

∆Q˙ _i = the heat flow transferred in the i^th pipe level,

k_i = heat transfer coefficient of the i^th pipe level,

A_i = heat transfer area of the i^th pipe level.

The mean driving temperature difference ∆t_αν,i of the i^th pipe level is defined as follows, see Figure 6:

∆t_αν,i = ——    1    2  · [(t_K,1 + t_K,2)_i – (t_WT,1 + t_WT,2)_i]⇥(5)

The calculations can either begin at the topmost pipe level, where t_K,1 = t_K,in and t_WT,2 = t_WT,out apply, or at the bottommost pipe level, with t_K,2 = t_K,out and t_WT,1 = t_WT,in. This iterative procedure can be briefly illustrated on the basis of a calculation commencing at the topmost pipe level, i = 1: using the given temperatures (t_K,1)₁ = t_K,in and (t_WT,2)₁ = t_WT,out, the required material data and the first values for k₁ and ∆t_αν,1 are determined. Equation (4) then utilizes the known exchange area A₁, see below, to deliver an estimated value for ∆Q˙ _i, which can be applied for calculating approximate values for (t_K,2)₁ and (t_WT,1)₁ using the local heat balances for clinker and heat transfer medium. These approximations form the basis for calculating new mean reference temperatures, which in turn are the basis for a further calculation. These iteration loops are run until a defined degree of accuracy is reached. The start values for calculating the next-lower pipe level, i = 2, are then: (t_K,1)₂ = (t_K,2)₁ and (t_WT,2)₂ = (t_WT,1)₁.

The given values for the clinker mass flow M˙_K, the bulk material density ρ_bed of the clinker bed, its speed of downward movement u_K, the cooling pipe length L_R and the cooling pipe geometry and arrangement unambiguously define the width of the g-cooler, the number of cooling pipes per cooling pipe level and thus also the heat transfer area A_i of a pipe level. The required total heat transfer area A_tot = ∑A_i must then be implemented by designing a corresponding number of levels = cooler height.

The calculation model was checked on the basis of already installed systems and a test-plant cooler, which resulted in a satisfactory correspondence with the measured values. The model was therefore implemented as computer program [12].

3 Results and discussion

The following section presents main results of the study, whose consequences are analyzed and evaluated.

3.1 Comparison of air cooling and heat-transfer-oil cooling

To what extent does the construction size of a g-cooler change if the heat removal system is altered from a large flow of cooling air with relatively small temperature increase to a heat-transfer-oil flow with a significantly higher temperature increase? The construction size of the cooler can be characterized by the respectively necessary heat transfer area. Taking the example of the reference case described in section 1: from

Q˙ _tot = (k · A · ∆t_αν)_air = (k · A · ∆t_αν)_WT⇥(6)

follows

————   A_WT    A_air  = ————   k_air     k_WT  · —————    ∆t_αν,air     ∆t_αν,WT⇥(7)

For the driving temperature difference ∆t_αν between the clinker bed and the cooling medium along the heat transfer area, the mean logarithmic temperature difference of the pure counterflow is applied as an approximation [6]. The previous air cooling can be described with the following operating values: t_air,in ~= 30 °C, constant heating up by ∆t_air,in/out ~= 100 °C, thus t_air,out ~= 130 °C; applying the clinker temperatures of the reference case, ∆t_αν,air ~= 155 °C is obtained; with the empirical value α_air ~= 3 · α_K,tot, equation (2) then provides the heat transfer coefficient k_air ~= 0.75 · α_K,tot. The following applies to the thermo-oil cooling: k_WT ~= α_K,tot, (see Fig. 5); ∆t_αν,WT is calculated with variable values for t_WT,out. Equation (7) then provides the relationship between the relative heat transfer area (A_WT/A_air) and the thermo-oil outlet temperature t_WT,out depicted in Figure 7.

Despite the higher heat transfer coefficient k_WT ~= 1.33 · k_air, at temperatures above t_WT,out ~= 200 °C the heat-transfer-oil cooling requires an increasingly larger heat transfer area A_WT than that needed for the air cooling. This is due to the fact that in line with an increasing outlet temperature t_WT,out there is a steady decrease in the driving temperature difference ∆t_αν,WT between the heat transferring and heat absorbing material flows. Independent of the process involved, the downstream power generating plants require heat-transfer-oil temperatures of t_WT,out ≥ 300 °C. To fulfil this requirement, the heat transfer area has to be increased to A_WT ≥ 1.5 · A_air compared to the current standard g-cooler.

3.2 Comparison of lens-shaped pipe and round pipe

For the reference case, Figure 8 shows the curves for clinker temperature, heat transfer medium temperature and transferred heat flow occurring along the height of the cooler at a given thermo-oil outlet temperature of t_WT,out = 320°C in a g-cooler equipped with lens-shaped pipes. The clinker bed has a downward-movement speed of u_K = 4.5 cm/min at a clinker bed density of ρ_bed = 1400 kg/m3. The coolant mass flow is M˙_WT = 22.44 kg/s. A total of Q˙ _tot ~= 17.14 MW is transferred from the clinker to the heat transfer oil via a heat transfer area of A_tot = 21315 m2. The external dimensions of the pipe bundle are: (B_R · L_R · H_R) ~= (24.0 m · 3.6 m · 27.5 m).

For the same operating case, Figure 9 compares the above results with those obtained if round DN 100 pipes are used for the cooling instead of the lens-shaped pipes. This measure reduces the required heat transfer area to A_tot = 18180 m2. One of the reasons for this is the lower contact time τ_K,w between clinker and pipe wall, whose positive influence outweighs the negative effect of the stagnant zones at the clinker-flow-impact and rear sides of the pipes. As the heat transfer area per DN 100 pipe is smaller than that of a lens-shaped pipe, a significantly higher number of the round pipes has to be installed in the bundle. Nevertheless, a cost analysis shows that it is still cheaper to use DN100 pipes. The staggered arrangement of the rows of DN 100 pipes is retained, but the spacing of the pipes is designed as sketched in Figure 10, in order to ensure an undisturbed flow of the bulk material. This results in pipe-bundle dimensions of (B_R · L_R · H_R) ~= (27.5 m · 3.6 m · 27.5 m). The considerations discussed in the following sections are based on the use of DN 100 pipes.

3.3 Other influencing variables/comparisons

In the reference case, reduction of the outlet temperature of the heat transfer medium to t_WT,out = 300 °C by correspondingly increasing the flow rate of the medium, with DN 100 pipes and u_K = 4.5 cm/min, results in a decrease in the required heat transfer area to A_tot = 14475 m2. Cause: an increase in the driving temperature difference ∆t_αν,WT.

For the same operating case, an increase in the downward movement speed of the bulk material from u_K = 3.5 cm/min to u_K = 5.5 cm/min results in a decrease in the required heat transfer area A_tot of only around 5.5 %, related to the value at u_K = 3.5 cm/min. At a given clinker throughput M˙_K, and in line with a rising u_K, the floor area of the cooler decreases while the height of the cooler increases, and vice versa.

Given the same inlet and outlet temperatures of the clinker, the required heat transfer area A_tot increases in line with the rising clinker mass flow M˙_K. At a constant u_K, this results in coolers with increasingly larger floor areas but approximately constant heights.

In an analogue manner, various other influencing variables, including clinker inlet temperature, clinker particle size and type of heat transfer medium, were analyzed.

3.4 Selection of the heat transfer oil

The outlet temperature t_WT,out determines the type of heat transfer oil to be used. Examples: the synthetic heat transfer oil THERMINOL 66 from Fragol GmbH & Co. KG [13] covers a working range of (0 - 355) °C; price: P_TH66 ~= 5.0 EUR/l. The mineral-oil-based heat transfer oil FRAGOLTHERM Q-32-N from the same supplier can be used for continuous operation in a temperature range of ((-12)  - 320)°C; price: P_Q32N ~= 1.2 EUR/l. The physical data of the heat transfer medium only have a slight influence on the heat transfer conditions, but have a significant effect on the capital investment and operating expenses. To safely allow for operating disturbances, e.g. a flush of material leaving the kiln, the maximum thermo-oil operating temperature should remain about 20°C below its permitted upper operating limit. For the reference case with t_WT,out = 320 °C, THERMINOL 66 can be used, while for the case with t_WT,out = 300 °C, FRAGOLTHERM Q-32-N is suitable. These preconditions result in the following capital costs for the initial filling of the system with heat transfer oil: outlet temperature t_WT,out = 320 °C —­­–> K_TH66 ~=  2.31 Mio. EUR, t_WT,out­ = 300 °C —­­–> K_Q32N ~=  0.44 Mio. EUR. The thermo-oil filling volumes are based solely on the internal volumes of the DN 100 pipes required for the heat transfer areas calculated in the above text. However, in practice higher filling volumes are required, because it is also necessary to fill the volumes of, for instance, the heat exchanger used for absorbing the heat, the storage tank etc., i.e. even greater cost differences occur.

The above considerations serve for orientation. They show clearly that some of the specific capital costs can be drastically reduced by limiting the outlet temperature of the heat transfer medium to t_WT,out = 300 °C. The savings result from the smaller size of the g-cooler building/the enclosed volume of the building, the reduced amount of required heat transfer oil, the possible usage of considerably cheaper types of heat transfer medium and the lower cost of peripheral equipment. There is no associated limitation with regard to the cooler’s combination with downstream power generating systems, e.g. ORC, Kalina etc.

3.5 Kiln flush

For the reference operating case, the following material flush data were defined on the basis of plant operators’ experience: clinker mass flow M˙_K   ^* = 1.3 · M˙_K= 6500 t/d for a period of ∆τ^* = 20 min, clinker temperature downstream of recuperator = g-cooler inlet temperature t^*_K,in = 584°C. This results in a flush material mass of ∆M^*_K = 81.25 t, which is deposited on/distributed over the g-cooler charge as a layer of ∆H^* ~= 0.6 m thickness. This flush material passes through the cooler in mechanically unmixed form. Due to the low thermal conductivity of the bed of material, only limited heat transmission to the surrounding colder clinker layers can take place. The actual cooling is carried out via contact with the cooling pipe surfaces. In this process the topmost rows of pipes are critical, because the layer of flush material passes uncooled into these. Various calculations show that this leads to ∆t^*_WT,out ~= 20 °C higher thermo-oil outlet temperatures than occur in steady-state conditions.

The steady-state operation at t_WT,out = 300 °C, discussed in section 3.3 as being the aim when using DN 100 pipes and FRAGOLTHERM Q-32-N, thus results in a maximum heat-transfer-medium temperature of t^*_WT,out ~= 320 °C when a flush of material leaves the kiln. This is only a temporary peak value at the upper operating limit of this heat transfer medium, and is tolerated by the manufacturer. However, during this temperature peak the so-called film temperature must not exceed a temperature of 340°C [13]. The kiln flush can only be indirectly simulated by means of the created computer program. Analysis results indicated that the outlet temperature of the hot layer of clinker would be t^*_K,out < 110 °C. Conclusion: a flush of material from the kiln can be safely handled by a g-cooler using heat transfer oil as the coolant.

3.6 Design modifications

The modified g-cooler is still made up of identical compartments arranged in so-called columns one above the other to suit the required bulk material residence time = bed height. The total heat transfer area A_tot results from the parallel connection of a number of these columns. The employed heat transfer medium flows through the individual columns as cross-counterflow to the bulk material. Each column is individually supplied with heat transfer oil independently of the other columns, i.e. a separate pump is allocated to each column. If problems occur, this system allows individual zones = columns of the g-cooler to be shut down, while emergency operation is maintained with the remaining active zones.

Design modifications are caused by the specific volume of heat transfer oil, which is about three powers of ten smaller than that of the originally employed cooling air. A drastic reduction is therefore necessary in the cross-section of the channels through which the heat transfer oil of a column flows. In order to produce a turbulent flow of oil, here: Re_WT ≥ 10000, u_WT ~= 1.0 m/s, it is necessary to combine (2 - 3) individual pipes to form one flow channel. This complicates the design of the return stations at the end faces of the compartments.

At the moment, studies are examining the impact on cooler construction size and costs of enlarging the surface at the bulk-material flow side of the cooling pipes by means of cooling fins aligned parallel to the clinker flow. In order not to cause flow problems in the bulk material, the cooling fins would have to be relatively far apart. This limits the additionally installable heat transfer area.

4 Alternative solution

An alternative process appears interesting in view of the size of the modified g-cooler in conjunction with the resultant substantial costs for heat transfer oil, as well as the expense of equipment for oil circulation between cooler and power generating plant, safety system, monitoring, general handling, emergency system for heat dissipation in the event of power generating plant failure etc. This is the use of ambient air as the cooling medium instead of heat transfer oil. This cooling air would be heated up to t_air,out ≥ 320 °C and supplied to the power generating plant in an open system, i.e. after heat transfer to the working medium of the power generating plant, the air would be discharged into the atmosphere or supplied to other plant components for utilization of the residual heat.

Advantages: the above-mentioned problems relating to the cost of heat transfer oil and its handling would be avoided. Moreover, the power generation plant and the g-cooler would be very extensively decoupled and the emergency heat-dissipation system would be greatly simplified. In principle, the emergency system could consist of a controlled discharge of heat/cooling air into the atmosphere.

Disadvantages: compared to a g-cooler using heat transfer oil, the g-cooler using cooling air would have to be larger. Using the approximation employed in section 3.1, it can be estimated that the heat transfer area A_air for air cooling would have to be approx. 33 % larger than the A_WT for oil cooling, i.e. A_air ~= 1.33 · A_WT. This applies to equal  in- and outlet temperatures of air and heat transfer oil. It is therefore also necessary to check the heat transfer area of the waste heat boiler of the power generating plant.

It is only possible to evaluate the two above-discussed alternative solutions on a holistic basis, taking consideration of the overall system of recuperator, g-cooler and power generating plant.