The invention relates to a spiral heat exchanger provided with a cylindrical, casing, inside which a first medium can flow through, and a channel which runs as a spiral around the cylinder axis and through which a second medium can flow. A spiral heat exchanger of this type is disclosed in Offenlegungsschrift DE-3519315 Al.

In the case of the heat exchanger disclosed in this publication, a channel in spiral form through which a second medium flows is provided inside the cylindrical casing through which a first medium flows. This heat exchanger is usually used in a central heating circuit in which the exhaust gases flow out of the heating boiler through the casing and in part release heat to the return water flowing through the spiral channel, which yields a saving in fuel consumption. The spiral channel is double- wound so that inlet and outlet are on one side of the cylindrical casing. A problem with this heat exchanger is the relatively moderate heat transfer.

In the domestic environment and in the process industry, heat exchangers of compact size have now been found to be of great importance, on the one hand in connection with the lack of space under certain process conditions and, on the other hand, in connection with the advantageous specific power (kW/m ³ ) of such a heat exchanger for a specific temperature difference between the media and the low specific cost price per square meter of heat-exchanging surface which is possible as a consequence. In addition in industry, applications have emerged for which it was found necessary to have a heat exchanger of small dimensions in which several media flows can exchange heat with one another.

The aim of the invention is to overcome the above-mentioned problems and to provide a compact heat exchanger in which the heat transfer is particularly high and with which, as a consequence of the dimensions, a modular construction of a plurality of these heat exchangers is simple to achieve.

This is achieved in the case of a heat exchanger of the type mentioned in the preamble in that, in addition to the one spiral channel and in each case alternating t erewith, at least one additional spiral channel is provided through which a third medium can flow, and in that a straight channel through which the first medium can flow is formed centrally within the two spiral channels.

As a result of this compact design according to the invention with the correspondingly small radius of curvature of the spiral channels, a

severe turbulence is produced in the media in these channels, as a result of which an excellent heat transfer is obtained. At the same time, the general requirement for heat exchangers that the surface area should be as large as possible with as small as possible a flow-through space is met. As a result of this design with three or more channels, several flows can enter into heat exchange with one another in various ways. By enclosing the straight channel within the two or more spiral channels, the hottest medium can be fed through the straight channel and the colder media to be heated can be fed through the two or more spirals. Consequently, the larger outer surface area remains relatively cold and it may be possible to operate without insulation or with limited insulation on the cylindrical outer side.

The invention also relates to a matrix-type heat exchanger built up from modules, each of which are formed by a heat exchanger as mentioned above.

The invention will be illustrated in more detail with the aid of an illustrative embodiment, with reference to the drawings, in which:

Fig. 1 gives a partially perspective and partially cross-sectional view of the heat exchanger according to the invention; Fig. 2 gives a graph of the influence of the spiral diameter and the water speed on the heat transfer coefficient; and

Fig. 3 gives a graph of the medium temperature in the channels as a function of the spiral length.

Figure 1 gives a view of the compact exchanger 3 according to the invention with two spiral channels or pipes 1, 2, each of which alternately is wound directly around the straight cylindrical channel 4. As a result of the very compact construction with an outside diameter of about 19 mm and a diameter of each channel of approximately 6 mm, a severe turbulence of the media, and consequently an excellent heat transfer, is obtained due to the very small radius of curvature of the spiral channels and also due to the walls which have very good thermal conductivity. The cylindrical casing of the heat exchanger 6 can be made of insulating material. In one embodiment, each spiral pipe has sixteen windings around the straight pipe, as a result of which the total heat exchanger has a length of about 235 *•*••*-• The heat exchanger according to the invention can advantageously be produced, for example, by winding two annealed red copper pipes, filled with sand, with a guide around a steel pin to form an assembly. Each copper pipe, for example, has an inner(di)- and outer(du) diameter of 4.6 mm and 6.35 ^mm respectively. After removal

of the pin, a straight red copper pipe having the same external diameter is inserted in the same place. The entire assembly is then immersed in liquid tin or another liquid materials aving good thermal conductivity. In order to provide optimum contact between the tin and the copper pipes and to fill all spaces between the pipe walls, the tin is subsequently melted again using a burner flame. By this means, any hollow cavities still present between the three pipes are completely filled by the liquid metal flowing in until full.

It is self evident that other means of production and also other materials can be used, depending on the media flowing through and on the applications. In this context, it is possible to proceed in such a way that, by production of the two spiral channels, which are provided with their own wall, and the filling material introduced in these, the straight central channel is simultaneously formed without its own specific wall. The straight channel then acquires a corrugated wall formed by the spiral walls and filling material.

It is also particularly advantageous to bend the inlet and outlet ends of the spirals in such a way that they are parallel to the inlet and outlet end of the straight channel. In this way, a number of these heat exchangers can be assembled in modular construction to form a larger matrix-type heat exchange unit which will be illustrated below.

Finally, polyethylene can be applied as pipe insulation 6 around the entire assembly.

A few trials and the measurement results of these will be explained below, on the basis of which it can be deduced that, with this heat exchanger using water/water as media, a specific power of at least 21000 kW/m ³ can be achieved with an average temperature difference between the media of 43. ^β C. Under these conditions, a k-value of 9200 W/m.K and a heat transfer coefficient of about 2300 W/m ² .K are achieved. These values are high compared with those of the heat exchangers from the current state of the art.

The following data can also be derived from the values ind ⁱ cated in Tables 1 and 2 below for the case of water/water (Tab? I) and air/nit- ?gen vapour (Table 2) flowing through two or three pipes. - maximum k-value of 9200 W/m ² .K (water/water) and 10b W/m ² .K (gas/gas) : maximum specific power of about 21000 kW/m ³ (water/water) and 280 kW/m ³ (gas/gas); and maximum heat transfer coefficient α of 22600 W/m ² .K (water/water)

and 210 W/m ² .K (gas/gas) .

It is readily conceivable that even higher values are achieved in the channels at water speeds which are higher than the speed of 2.5 m/s used for the above-mentioned trials. It is self evident that in the case of the above-mentioned heat exchanger according to the invention it is of great importance that the material providing contact, that is to say the substance poured between the pipes, for example tin, has very good thermal conductivity between the three channels. The calculation of heat transfer coefficients on the basis of the measurement results is made more difficult because the wall temperatures of the channels through which there is flow were not measured. Since these trials were intended to obtain a global evaluation of this type of heat exchanger, it was elected to calculate the heat transfer coefficients making the following assumptions: the thickness of the tin layer between the various respective channels is a minimum of 1 mm at the locations where the copper channel walls are closest to one another; the tin melted between the channels has at all locations a 100.5! contact with the wall surfaces of the spirals and the straight channel; a value of 349 W/m.K is taken for the thermal conductivity coefficient of copper and a value of 65 W/m.K for that of tin; the arithmetic average value between the measured inlet and outlet temperatures of the two heat-exchanging media is taken for the temperature in the middle of the 1-mm thick tin layer (see also

Fig.3). It was assumed here that the change in temperature is linear in the longitudinal direction of the various channels.

The heat flow density through the channel walls was calculated with the aid of the transferred power, which follows from the heat balance and the internal surface area of the various channels. The heat transfer coefficient follows from the thermal resistance of half the material thickness (=■ half tin layer thickness + one copper channel wall thickness), the calculated heat flow density and the average inside wall temperature of the channels. A calculation example is given below, while all calculated heat transfer coefficients for the media water/water (2130 to 22624 W/m ² .K) are given in Table 1.

The substantial difference in the calculated heat transfer resistances results because the internal surface area of the two spirals

together is a good six times as large as that of the straight cylindrical interior channel. Moreover, as a consequence of the operating conditions cho ' , the at loads during experiments Nos. 1 and probably differ by about a fac^r of two. The low heat transfer coefficients apply to the spirals and the high heat transfer coefficients to the straight cylindrical channel. The measurement results are used as inp- for a calculation model.

Table 1 shows the measurement results for the media water/water and Table 2 the measurement results for the media air/nitrogen vapour. The trials were carried out in order to gain an impression of how the heat exchanger behaves, both with liquid media and with gaseous media. The results were, as stated above, excellent.

Taking as a basis heat transfer for a turbulent flow in the spirals having Reynolds numbers of > 22000, as in the case of Experiments 1, 2 and 4 (see Table 1), a calculation of two heat transfer coefficients, both for the media water and air is given below.

The equations disclosed in VDI-Waermeatlas [VDI heat atlas] (1988), 5th edition, were used for this calculation. These equations are:

(€/8).Re.Pr.(Pr) ⁰ - ¹⁴

Nu=

1+12.7, ^~ Fe/κT. (Pr ^{2 3} -l ) . (Prw) ^0Λk

6= 0.3164/Re ⁰ - ²⁵ + 0.03.(d/D) ⁰ - ⁵

where: d - internal pipe diameter (di in Fig. 1) in m; D = spiral diameter (Dw in Fig. 1) in m;

Re = Reynolds number;

Pr = Prandtl number for the relevant medium;

Prw = Prandtl number for the relevant medium at the wall temperature prevailing there. Applied to Experiment No. 1, this gives the following:

e= 0.3l6^/2l6l6°- ²⁵ +0.03.(0.0046/0.0128) ⁰ - ⁵ = 0.044

(0.044/8).2l6l6.3.0.(3.0) ⁰ - ^{1 Nu= l87*6}

or α= Nu-λ/d=(l87.6.0.65)/0.0046= 26504 W/m.K

For hot air, Experiment No. 5. we find α = 177 W/m ² .K according to the above calculation example.

The influence of the spiral diameter and of the water speed on the calculated heat transfer coefficient is shown in Fig. 2. With a smaller spiral- or "coil" diameter Dw, such as used, a higher heat transfer coefficient α, and thus a higher heat transfer, is obtained for a certain water velocity.

With regard to Experiment No. 1, a simplified calculation of the heat transfer coefficient is now given, on the basis of the measurement results.

In this experiment water flowed through the two spirals in counter flow. The straight cylindrical channel was not used. The two ends thereof were not closed off, so that some natural convection of the ambient air may have occurred in this channel. The temperatures given in Table 1 were measured after the heat exchange was clearly in the steady state (with the exception of Experiment No. 5) • The outside of the two spirals was insulated with a piece of pipe insulation (polyethylene foam 13 mm thick). The logarithmic average temperature of 4θ.53 ^* C follows from the various temperatures measured at the inlet end and outlet end of the spirals. If the arithmetic average temperature difference is used, instead of the logarithmic value, this is then: ΔT av.(arith.)= ((75-1-40.0)+(59-2-12.7) ) /2~~ 4θ.8θ ^* C. ^' Because the deviation relative to the logarithmic average temperature difference is only 0.67#. the arithmetic average temperature difference is used below in this calculation. The change in temperature in the longitudinal direction of the heat exchanger is regarded as linear, as indicated in Fig. 3« This is permissible as a consequence of the very short dwell time of the water in the spirals (about 0.32 to about O.56 sec). Moreover, it can be assumed that the average temperature difference prevails at half the length of the spiral. The average temperature pattern in the spirals is then as follows: hot medium: (75-1+59.2)/2 = 67.15'C; " cold medium: (40.0+12.7)/2 = 26.35"C.

The average "tin layer" temperature t _aw-t can then be calculated from these two temperatures as: t _av . _t *-= (67.15+26.35)/2 = 46.75"C. This is an approximation of the actual conditions because the heat transfer coefficients in the two channels do not have to have the same values at

all locations. The thermal resistances can also show local differences as a consequence thereof.

The heat flow density q is calculated from the average power transferred (see Table 1, Experiment No. 1) and the internal surface area of the spirals (•= 0.00923 m ² ): q * 2200/0.00923 ■ 238353 W/m ² . The thermal resistance δ/λ of the total material thickness (copper + tin) was also calculated. This δ/λ was 20.4.10 ^"6 m ² .K/W. The temperature difference over the material thickness δ then follows from the average heat flow density q (238353 W/m ² ) and the thermal resistance δ/λ: ΔT(mat. hick.) = 238353-20.4.10" ⁶ = 4.86'C. From this temperature difference and the average "tin layer" temperature (t _av . _t ), it is then possible approximately to calculate the average internal surface temperatures of the two spirals. These are then: 46.75+4.86/2 = 49.18*C (hot-medium side) and 46.75-4.86/2 = 44.32'C (cold-medium side). Finally, the desired heat transfer coefficients then follow from the average heat flow density divided by the temperature difference (i.e. q/ΔT) between medium and wall, that is to say: at the hot side: 238353/(67-15-49.18) = 13264 W/m ² .K at the cold side: 238353/(44.32-26.35) ■ 13264 W/m ² .K In this case these heat transfer coefficients are identical. However, as a consequence of the difference in internal surface area between the interior straight channel and a spiral, the heat flow densities are not identical if the media flow through these channels! Therefore it is not entirely correct to calculate in the same way in this situation. Nevertheless, this inaccuracy has been accepted for the calculation of the heat transfer coefficients listed in Table 1 because the wall temperature corrections for these are only of small magnitude.

Using the heat transfer coefficients calculated up to now, it is now possible, as a "check" also, to carry out a check calculation, that is to say using the known relationship for calculation of heat loss of cylindrical pipes. For this case, the relationship is as follows: π.(Ti-T2)

Qcyl.= l/(αi.Di)+(l/(2.λcu)).ln(Dl/Di)+(l/2.λsn)).ln(D2/Dl) wherein:

Qcyl. =the transferred power in one channel (W/m); Ti =average temperature of the hot medium (*C);

T2 =average partition (tin layer) temperature of the hot medium (°C); C. _J =the heat transfer coefficient (W/m ² .K) calculated above;

λcu and λsn are the coefficients of thermal conductivity of copper and tin respectively at the prevailing temperatures (in W/m.K) and Di, Dl and D2 are the respective diameters associated with the above-mentioned symbols (in m) . Expressed in figures, this gives the following equation:

_n , π.(67.15-46.75)

Qcyl.=

1/13264.0.Oθ46 ⁺ (l/(2.349)).ln(6.35/4.6) ⁺ (l/(2.65)).ln(7.35/6.35) or Qcyl.= 3565 W/m

The unwound length of one spiral of the heat exchanger is about 0.643 m, which signifies that Q = 0.643.3565/1000 = 2.292 kW is relinquished by one spiral to the other. This value is in good agreement with the measured values (see Table 1, Experiment No. 1) of 2.21 and 2.19 kW. Similar check calculations are, of course, also possible for the other experiments.

The excellent heat transfer of the spiral heat exchanger under consideration can be seen from the measurement (see Tables) and calculation data explained above. In the case of this heat exchanger according to the invention several media flows can advantageously exchange heat with one another at the same time. Using said one straight channel and the two or more spiral channels wound about the latter, several combinations are possible with respect to the media flows subjected to heat exchange. This gives a great flexibility with regard to these flow-through media.

As stated above, as a consequence of the relatively small dimensions, a number of these heat exchangers can advantageously be assembled in modular construction to form a larger matrix-type heat exchange unit. In a matrix-type heat exchange unit of this type a number of modules, each of which comprises a heat exchanger according to the invention, are joined to one another in rows and/or columns and connected to one another. A heat exchange unit of this type can be produced, for example, by winding lead wires in a block of ceramic powder in accordance with the said channels of a single heat exchanger according to the invention. The said block of ceramic powder is subsequently baked, the lead wires melting away in each module. After cooling, the said three or more channels remain in the ceramic material in each module of the exchange unit. A heat exchange unit of this type can advantageously be used for applications using flow-through media for elevated temperatures, for example above 1000 ^β C.