This invention relates to a control system and a control method for controlling a chemical reactor.

A plant and process are described in WO 2005/10251 1 (GTL Microsystems AG) in which methane is reacted with steam, to generate carbon monoxide and hydrogen in a first catalytic reactor; the resulting gas mixture is then used to perform Fischer-Tropsch synthesis in a second catalytic reactor. The reforming reaction is typically carried out at a temperature of about 800 °C, and the heat required may be provided by catalytic combustion in channels adjacent to those in which reforming is carried out, the

combustion channels containing a catalyst which may comprise palladium or

palladium/platinum on an alumina support in the form of a thin coating on a metallic substrate. An inflammable gas mixture such as a mixture of methane and air is supplied to the combustion channels. Combustion occurs at the surface of the catalyst without a flame. WO 2009/101434 (CompactGTL pic) describes a module for performing this steam methane reforming reaction in two stages.

Chemical reactors that define first and second flow paths in proximity, so there is good heat transfer between the different channels, may be used for a variety of different endothermic reactions, using appropriate catalysts. In each case the heat may be provided by an exothermic reaction such as catalytic combustion in channels adjacent to those in which the endothermic reaction is performed. A potential problem with such reactors is a thermal runaway, which occurs if the rate at which heat is being generated by the exothermic reaction exceeds the rate at which it is absorbed by the endothermic reaction. The temperature will rise, which may cause the rate of the exothermic reaction to increase, so the temperature rises even more, and the rate of increase may become exponential. In any such chemical plant it is therefore essential to control the processes taking place in such a way that operation of the plant is stable. With a laboratory-scale reactor it may be sufficient for the operator to observe closely, and to adjust flow rates manually, but for larger-scale reactor systems with multiple reactors and small buffer volumes an automated control system would be advantageous.

According to the present invention there is provided a control system for a chemical reactor in which an exothermic reaction takes place, the control system comprising a sensor arranged to monitor a parameter indicative of temperature within the chemical reactor, a rate-of-change calculator to determine the rate of change of temperature with time, at least one rate-monitoring means to monitor the observed rate of change of temperature and to provide an output signal in response to the observed rate of change, and flow control means responsive to the output signal to control the flow rate of reactants to the reactor. The sensor may be located within the reactor, or outside the reactor, and it may measure temperature, or another parameter which can be related to the temperature within the reactor. The sensor may be for example a thermocouple or a thermistor arranged to sense the temperature of gases flowing out from the reactor, or alternatively the sensor may be for example a thermocouple arranged to sense the temperature of gases within a flow channel of the reactor, which may be a flow channel for the exothermic reaction; or alternatively the sensor may be arranged to monitor the temperature of a wall defining a flow channel within the reactor.

The flow control means may control the flow rate of reactants for the exothermic reaction, and may decrease the flow rate by a predetermined amount, or an amount preset during commissioning, or by a calculated amount.

The control system may comprise at least three temperature sensors, and a voting means associated with the temperature sensors, to provide a temperature-representative signal. This ensures that failure of a temperature sensor does not lead to the plant changing its operating conditions, or even closing down. There may be several temperature sensors, for example more than ten, and in this case the voting means may compare all the sensed values, excluding any that appear to be incorrect, and calculating a representative temperature from the remaining values.

The monitoring means may comprise a comparison means to compare the observed rate of change to a threshold, and to provide the output signal if the threshold is exceeded. The control system may comprise at least two such comparison means, with different thresholds, which provide different output signals. Furthermore, if the flow rate is decreased to a lower value, it may be held at the lower value for a period of time, and the duration for which it is held at the lower value may differ. Hence, if the rate of change exceeds a first threshold, the flow control decreases the flow rate by a first amount, and holds the flow rate at this decreased value for a first predetermined period of time, while if the rate of change exceeds a second threshold, the flow control decreases the flow rate by a second amount, and holds the flow rate at this decreased value for a second predetermined period of time. Each threshold value may be adjustable, depending on the current state of the reactor. For example each threshold value during start-up may be greater than the corresponding threshold values during steady-state or on-going operation. The control system may also comprise a temperature-monitoring means which is supplied with a signal representing the observed temperature in the reactor, the temperature-monitoring means providing an output signal in response to the observed temperature. For example the temperature-monitoring means may compare the observed temperature in the reactor with a temperature threshold, and if the temperature threshold is exceeded the output signal is provided so that the flow control alters the flow rate. The signal representing the observed temperature may be the signal provided by the voting means.

It has been found by experiment that the most significant parameters for detecting thermal runaway are the rates of change of temperature within the reactor, the

temperature in the reactor, and the temperature differences in the reactor for example between inlet and outlet, or between the highest and lowest value. The thermal runaway may be controlled by changing either the composition of the gas mixture, or the flow rate of the gas mixture, supplied to the reactor. In many cases it is not practicable to adjust the gas composition, so the flow rate of the gas mixture may be the only parameter which is controlled. As explained above the present invention involves means to control the flow rate of reactants to the reactor; that may involve controlling the flow rate of one or more reactants individually or jointly, or it may involve controlling the flow rate of the gas mixture.

Where a plurality of temperature sensors are connected to a voting means, the sensors may be arranged at positions within the reactor where the temperatures can be expected to remain similar to each other. For example it may not be appropriate to vote on signals representing temperatures sensed or inferred at opposite ends of a reaction channel, as there may be a significant temperature difference between the opposite ends, whereas signals representing temperatures sensed or inferred at the same distance into parallel channels for the exothermic reaction would be expected to remain similar to each other. Where there are a larger number of temperature sensors, they may be distributed throughout the reactor to provide an indication of the temperature throughout the reactor.

The output signal which is generated in response to the observed rate of change of temperature may be used to adjust a flow rate set point, this set point being supplied to a signal generating circuit which generates a control signal, and the control signal being used to control the flow rate of reactants to the reactor.

By way of example such a control system may comprise a ramp unit to which is provided data representing the current gas flow and a desired gas flow set point, the ramp unit providing a control signal to a control unit, wherein the control unit also receives data representing the desired gas flow set point. The control unit, in response to the control signal from the ramp unit and to the data representing the desired gas flow set point may either provide a signal to adjust the gas flow, or may provide a signal to adjust the desired gas flow set point.

In the case in which the control unit provides a signal to adjust the gas flow, a signal to adjust the desired gas flow set point may be derived from an observed temperature, and a temperature set point. In the case in which the control unit provides a signal to adjust the desired gas flow set point, the ramp unit may additionally be provided with data representing an observed temperature and a temperature set point.

In a second aspect, the present invention provides a method of controlling a chemical reactor in which an exothermic reaction takes place.

The reactor may be a compact catalytic reactor in the form of a reactor block defining a multiplicity of first and second flow channels arranged alternately within the block to ensure thermal contact between the first and second flow channels. The flow channels in such a reactor block may be of length at least 300 mm, for example at least 500 mm, but typically no longer than 1000 mm. A suitable length is between 500 mm and 800 mm, for example 600 mm.

As indicated above, the invention is applicable in contexts in which the exothermic reaction is combustion. For example it is applicable to a steam/methane reforming reactor in which heat is provided by catalytic combustion of a combustible gas mixture which comprises a fuel (such as methane) and a source of oxygen (such as air). The

combustible gas mixture may be preheated to an elevated temperature below its auto- ignition temperature, before being fed into the reactor, and the steam/methane mixture may also be preheated. For example, all gases may be preheated to at least 500 °C.

Within a reactor the first flow channels and the second flow channels may extend in parallel directions, and the combustible gas mixture and the steam reforming mixture may flow in the same direction (co-flow) during operation. It has been found that co-flow operation gives better temperature control, and less risk of hot-spots.

In the reactor, each channel which is intended for a chemical reaction must contain an appropriate catalyst. The catalyst may be removable, and may be in the form of a catalyst structure. Each catalyst structure may comprise a metal substrate, and incorporate an appropriate catalytic material. Each catalyst structure may include a ceramic support material on the metal substrate, which provides a support for the catalyst. Each such catalyst structure may be shaped so as to subdivide the flow channel into a multiplicity of parallel flow sub-channels. Alternatively, the catalyst may be on the surface of the channel walls, or in the form of pellets, particles, particulates, powders, extrudates or spheres packed within the channel.

The metal substrate provides strength to the catalyst structure and enhances thermal transfer by conduction. The metal substrate may be of a steel alloy that forms an adherent surface coating of aluminium oxide when heated, for example a ferritic steel alloy that incorporates aluminium (eg Fecralloy (TM)). The substrate may be a foil, a wire mesh or a felt sheet, which may be corrugated, dimpled or pleated; a suitable substrate is a thin metal foil for example of thickness typically between 50 μηι and 200 μηι, for example 100 μηι, which is corrugated to define the longitudinal sub-channels.

Each reactor may comprise a stack of plates. For example, the flow channels may be defined by thin metal sheets that are castellated and stacked alternately with flat sheets, the plates being stacked and then bonded together; the edges of the flow channels may be defined by sealing strips. Alternatively the first and second flow channels may be defined by grooves in respective plates, which are then stacked and bonded together. To ensure the required good thermal contact both the first and the second gas flow channels may be between 10 mm and 2 mm high (in cross-section); and each channel may be of width between about 3 mm and 25 mm. The stack of plates forming the reactor block is bonded together for example by diffusion bonding, brazing, or hot isostatic pressing.

The invention will now be further and more particularly described, by way of example only, and with reference to the accompanying drawings, in which:

Figure 1 shows a diagrammatic side view of a reaction module for steam/methane reforming, incorporating a controller in accordance with the present invention;

Figure 2 shows a modification to the reaction module of figure 1 ; Figure 3 shows graphically the variation of temperature through the reactor module of figure 1 , and the corresponding variation of conversion in the steam methane reaction; Figure 4 shows a block diagram for a controller of the reaction module of figure 1 or figure 2;

Figure 5 shows graphically the desired rate of change of temperature pre-set in the controller of figure 4;

Figure 6 shows graphically how the ramp signal in the controller of figure 4 may vary with time; and

Figure 7 shows a block diagram for an alternative controller suitable for use with the reaction module of figure 1 or figure 2.

Referring now to figure 1 there is shown a reaction module 10 suitable for use as a steam reforming reactor. The reaction module 10 consists of two reactor blocks 12a and 12b each of which consists of a stack of plates that are rectangular in plan view, each plate being of corrosion resistant high-temperature alloy. Flat plates are arranged alternately with castellated plates so as to define straight-through channels between opposite ends of the stack, each channel having an active part of length 600 mm. By way of illustration, the height of the castellations (typically in the range 2-10 mm) might be 3 mm in a first example, or might be 10 mm in a second example, while the wavelength of the castellations might be such that successive ligaments are 20 mm apart in the first example or might be 3 mm apart in the second example. All the channels extend parallel to each other, there being headers so that a steam/methane mixture can be provided to a first set of channels 15 and an air/methane mixture provided to a second set of channels 16, the first and the second channels alternating in the stack (the channels 15 and 16 being represented diagrammatically), such that the top and bottom channels in the stack are both combustion channels 16. Appropriate catalysts for the respective reactions are provided on corrugated foils (not shown) in the active parts of the channels 15 and 16, so that the void fraction is about 0.9. A flame arrestor 17 is provided at the inlet of each of the combustion channels 16.

By way of example there may be over fifty such castellated plates in each stack. Each castellated plate defines several side-by-side channels 15 or 16; if the plate is of width 300 mm and each channel is of width 20 mm, there may be more than a dozen side- by-side channels 15 or 16 in each plate (allowing for the width of any sealing strips along the sides).

The steam/methane mixture is supplied through an inlet flow controller 30, and flows through the reactor blocks 12a and 12b in series, there being a duct 20 connecting the outlet from the channels 15 of the first reactor block 12a to the inlet of the channels 15 of the second reactor block 12b. Similarly the combustion mixture is supplied through an inlet flow controller 31 , and also flows through the reactor blocks 12a and 12b in series, there being a duct 22 connecting the outlet from the channels 16 of the first reactor block 12a to the inlet of the channels 16 of the second reactor block 12b. The duct 22 includes an inlet 24 for additional air, with an inlet flow controller 32, followed by a static mixer 25, and then an inlet 26 for additional fuel with an inlet flow controller 33, followed by another static mixer 27. Inlet flow controllers 30, 31 and 33 are implemented as control valves that control the flow of the steam/methane mixture, combustion mixture and additional fuel respectively. Each reactor block 12a and 12b is provided with a controller 35, whose operation is described below.

In use of the reaction module 10, the steam/methane mixture is preheated to 620 °C, and supplied to the reaction module 10 to flow through the reactor blocks 12a and 12b. A mixture of 80% of the required air and 60% of the required methane (as fuel) is preheated to 550 °C, which is below the auto-ignition temperature for this composition, and is supplied to the first reactor block 12a. In both cases the preheating may be carried out by heat exchange with exhaust gases that have undergone combustion within the module 10. The temperature rises as a result of combustion at the catalyst, and the gases that result from this combustion emerge at a temperature of about 700 °C. They are mixed with the remaining 20% of the required air (by the inlet 24 and the static mixer 25), and then with the remaining 40% of the required methane (by the inlet 26 and the static mixer 27), so that the gas mixture supplied to the combustion channels 16 of the second reactor block 12b is at about 600 °C, which is again below the auto-ignition temperature for this mixture (which contains water vapour and carbon dioxide as a consequence of the first stage combustion). By adjusting the temperature of the additional air supplied at the inlet 24, the temperature of the resulting mixture can be controlled to be below the auto-ignition temperature. Referring now to figure 2 there is shown a reaction module 100 which is a modification of the reaction module 10. The reaction module 100 has a single reactor block 12, equivalent to the first reactor block 12a, and the duct 20 connected to the outlet from the channels 15 of the reactor block 12 therefore constitutes the outlet for the synthesis gas. Hence the reaction takes place in a single stage, but in all other respects the reaction module 100 is as described above.

Referring now to figure 3, this shows graphically the variations in temperature T along the length L of the combustion channels 16 (marked A), and that along the reforming channels 15 (marked B) in the reaction module 10 of figure 1 . The portion of the graph between L = 0 and L = 0.6 m corresponds to the first reactor block 12a, while the portion of the graph between L = 0.6 m and L = 1 .2 m corresponds to the second reactor block 12b. It will be noted that the temperature T in a reforming channel 15, once combustion has commenced, is always lower than the temperature T in the adjacent combustion channel 16. The combustion gas temperature undergoes a downward step change as a result of the added air (from inlet 24) between the first reactor block 12a and the second reactor block 12b (at position L = 0.6 m). The variation of conversion of methane, C, in the steam reforming reaction with length L is shown by the graph marked P. The conversion increases continuously through the reaction module 10 and reaches a value of about 80%, which is close to the equilibrium conversion under the reaction conditions.

The total quantity of fuel (e.g. methane, natural gas; methanol, ethanol or other oxygenate-based fuel; or a fuel containing hydrogen) that is required is that needed to provide the heat for the endothermic reaction, and for the temperature increase of the gases (sensible heat), and for any heat loss to the environment. The quantity of air required is up to 10% more than that needed to react with that amount of fuel. The above summarises the mode of operation when a steady-state has been achieved. During start-up, the flow rate of methane, or of a methane/air mixture, must be gradually increased until the desired steady-state is achieved.

For example, each reactor block 12a, 12b may be initially brought up to 550 °C by using electrical heaters, or using a fuel specifically provided for start-up, such as methanol, ethanol or a mixture of oxygenates; the blocks 12a and 12b may be heated directly, or by heating an inert gas such as nitrogen and passing it through the channels 15 and 16. When this temperature has been reached, the steam and methane mixture is preheated in the same way and supplied into the reactor module, and the flow rate is raised to the desired value. The flow rate of this steam/methane mixture may be increased in steps. Then the combustible gas mixture, for example air and methane, is supplied to the combustion channels 16. Where a different fuel, such as an oxygenate, is used during start-up, the proportion of the start-up fuel may be decreased as the proportion of the steady-state fuel (eg methane) is increased; and in each case the changes may be continuous or step-wise.

The flow rate of the combustible gas mixture is then slowly increased so the temperatures of the reactor blocks 12a and 12b increase to the operating values. Initially the flow rate of the mixture supplied to the first reactor block 12a, and the flow rates of the air 24 and additional fuel 26 supplied to the second reactor block 12b, can be increased rapidly, but as the temperatures increase and approach the normal operating values, the flow rates must be increased more slowly. It has been found that the last 15% of the desired fuel composition is the most critical, being the situation in which a thermal runaway is most likely to occur. It has been found that in a thermal runaway the temperature can increase very rapidly, for example at over 20 ^q C/s, and that the increase may be exponential; such a rapid temperature increase would be very difficult to control by manual intervention.

For a two stage reactor configuration, such as that shown in Figure 1 , the fuel percentage introduced into the second stage reactor block 12b remains lower than that introduced into the first stage reactor block 12a throughout the start-up procedure. This results in the first stage reactor block 12a reaching steady state operating conditions in advance of the second stage reactor block 12b.

Referring now to figure 4, there is shown a block diagram of a control system which has been developed for laboratory systems where the scientists control some experiments based on flow rate as the governing factor, rather than the maximum temperature. Figure 4 shows a block diagram of the controller 35 of the first reactor block 12a, which controls the inlet flow controller 31 . The controller 35 requires information on the process variables, which are: the rate of flow of methane/air mixture into the channels 16 of the reactor block 12a; the gas composition entering these channels 16; and the temperature within the first reactor block 12a. These data inputs are indicated at 36, 37 and 38 respectively. The rate of flow may be measured using a coriolis mass flow meter, so it represents the mass flow rate. The gas composition data enables the calorific value of the gas or gas mixture to be deduced, and in this controller 35 the data signal 37 indicates that calorific value. In the case of the temperature data input 38, this is preferably derived from several temperature sensors such as the thermocouples 39 (only one is shown in reactor block 12a, but three are shown in reactor block 12b of figure 1 ), typically at least ten, distributed through the reactor block 12a, the resulting temperature measurements being subjected to a voting process 40 (see Figure 1 ) in which any values which are very different from the others are excluded, and the remaining values are used to provide the data input 38, for example by averaging.

The observed gas flow rate data 36 is supplied to a ramp unit 42 to which is also supplied a signal representing a flow set point 44 which may be adjusted in accordance with the observed temperature 38 (as described below). The ramp unit 42 is also provided with the observed temperature data 38. The ramp unit 42 provides a ramp signal 45 indicating a currently-desired temperature; during start-up the currently-desired temperature gradually increases for example initially at 1 °C/m n, but the rate of increase gradually decreases to zero as the temperature reaches the optimum reactor operating temperature. This ramp signal 45 is provided as an input to a control loop feedback controller 46, for which it provides a set point. The control loop feedback controller 46 is also provided with the observed flow data 36, and the gas composition data 37. The gas composition data 37 in this case indicates the calorific value of the gas, and provides a scaling factor for the ramp signal 45. The control loop feedback controller 46 compares the scaled ramp signal to the observed flow data 36, and provides a control signal 48 to adjust the inlet flow controller 31 .

The ramp unit 42 is also provided with a control input terminal 43 whereby its operation may be modified, as described below.

In an alternative, the gas composition data 37 might instead be supplied to the ramp unit 42. In this case the gas composition data may comprise data on the

components of the gas mixture, from which the ramp unit 42 can calculate the calorific value. In this alternative, the ramp signal 45 provided by the ramp unit 42 would indicate the currently-desired flow rate. In this alternative, the control loop feedback controller 46 would only have the two inputs, the observed flow rate data 36 and the ramp signal 45.

The observed temperature data 38 is supplied as an input to a control loop feedback controller 50 to which is also supplied a temperature set point 52; this provides an output signal 54 which adjusts the flow set point 44 either up or down. During start-up, the flow set point 44 is therefore initially held at the value expected to achieve the pre-set temperature set point 52; however, if the observed temperature 38 exceeds the pre-set temperature set point 52, then the flow set point 44 would be decreased. Each control loop feedback controller 46, 50 may be a PID controller (Proportional,

Integral, Derivative controller), as is widely used in industrial control mechanisms, or may be a more complex electronic controller such as an internal model controller or an advanced process controller. It might comprise a microprocessor. The ramp unit 42 may be embodied as a set of rules within a Programmable Logic Controller (PLC) or a

Distributed Control System (DCS).

Referring now to figure 5, this shows graphically how the rate of change of temperature, dT/dt, is to vary with temperature, T. This relationship is pre-set in the ramp unit 42, so the ramp unit 42 can calculate an appropriate ramp signal 45 (which as mentioned above, represents a currently desired value of temperature), as the reactor block 12 is heated up to temperatures above 550 °C by performing combustion of fuel in the combustion channels 16, in the situation where the reactor block 12 had been heated up to 550°C using electrically-heated nitrogen, as described above. In this example the flow set point 44 corresponds to a reactor temperature of 750 °C. The value of dT/dt is initially set at 1 'C/rnin, and remains constant up to a temperature of 650 ^< €; once the reactor temperature exceeds 650 °C the rate of increase dT/dt gradually decreases, such that for a temperature of 750 °C the rate of increase dT/dt becomes zero. In this example the variation between 650 °C and the desired reactor temperature of 750 °C is linear, but it will be appreciated that a different functional relationship may be used.

Referring to figure 6, this shows graphically how the ramp signal 45 may therefore vary with time, as the temperature of the reactor is heated up to 750 °C. The graph S illustrates the ideal variation, with the temperature gradually increasing and stabilising at the desired value.

Referring again to figure 4, the observed temperature data 38 is also supplied to a rate-of-change unit 56 which provides an output signal 58 corresponding to the rate of change of observed temperature with time. The input values are smoothed to remove noise before making the calculation. The rate-of-change unit 56 is implemented as a combination of hardware and software on either a PLC or a DCS as used in the art of industrial process control. The signal 58 is provided as an input to two comparators 60 and 62. The first comparator 60 is also provided with a first rate-of-change threshold 61 , while the second comparator 62 is also provided with a second rate-of-change threshold 63, the second rate-of-change threshold 63 being larger than the first rate-of-change threshold 61 . For example the threshold 61 may be 0.5°C/s, while the threshold 63 may be 2 ^q C/s. These threshold values are significantly greater than the preferred rate of temperature increase, which in this example is no greater than 1 °C/min, but are sufficiently high that short-term temperature fluctuations are unlikely to trigger the comparators 60 or 62, and are nevertheless significantly less than the rates of change that would occur in a thermal runaway. The values of the thresholds 61 and 63 given here are by way of example only; and it will also be appreciated that the values of the thresholds 61 and 63 may be reduced once steady-state operation has been achieved. If the output signal 58 exceeds the first rate-of-change threshold 61 , a control signal as indicated at boxes 64 and 66 is sent to the terminal 43 of the ramp unit 42. This causes the ramp unit 42 to decrease the ramp signal 45 by a first predetermined amount, and to impose a pause on any further increase of the ramp signal 45 during a time delay. This time delay may be pre-set for example at 5 minutes or 10 minutes, or alternatively may be until the observed temperature data 38 has stabilised. Similarly, if the output signal also exceeds the second rate-of-change threshold 63, then a control signal as indicated at boxes 66 and 68 is sent to the terminal 43 of the ramp unit 42. This causes the ramp unit 42 to decrease the ramp signal 45 by a second predetermined amount, and to impose a pause on any further increase of the ramp signal 45 during the time delay. The second predetermined amount is greater than the first predetermined amount, so that the ramp signal 45 is decreased still further, which as explained above lowers the effective set point for the control loop feedback controller 46 still more, so that the flow controller 33 reduces the gas flow rate by a larger amount.

Figure 6 shows, as graph R, such a variation of the ramp signal 45: in this example the temperature increases along the desired temperature graph S up to about 692 °C, but at that time the observed temperature data 38 indicated that the temperature was rising too rapidly, so the ramp signal 45 was dropped to correspond to a temperature of 680 °C, held at this value for a time delay, and then gradually increased again.

Such a rapid temperature increase may occur more than once; the graph Q illustrates a further variation of ramp signal: the temperature was observed to be rising too rapidly at a temperature of 730 °C, so the ramp signal 45 was dropped to correspond to a temperature of 720 °C, held at this value for a time delay, and then gradually increased again.

It will be understood that the controller 35 of figure 4 monitors both the

temperature and the rate of change of temperature throughout operation. So, during the ramp-up mode (during start-up), not only does the controller 35 reduce the fuel flow if the observed temperature 38 exceeds the temperature set point 52, but the controller 35 will also respond to a high rate of change of temperature. If the rate of change of temperature is above a threshold, a control signal as indicated at boxes 66 and 68 is sent to the terminal 43 of the ramp unit 42. This causes the ramp unit 42 to impose a pause on any further increase of the ramp signal 45 during a time delay, and, depending on the magnitude of the rate of change of temperature, it may also decrease the ramp signal 45. Referring now to figure 7 there is shown a block diagram of an alternative controller 350 for the first reactor block 12a, which controls the inlet flow controller 31 , and would take the place of the controller 35. It could also be used to control a single reactor block 12, as in the reactor module 100. The controller 350 is configured to be suitable for use in commercial operations where the maximum temperature is the most significant parameter, but in many respects has features in common with the controller 35; those features that are the same are referred to by the same reference numerals. The controller 350 requires information on the process variables, which are: the rate of flow of methane/air mixture into the channels 16 of the reactor block 12a; the gas composition entering these channels 16; and the temperature within the first reactor block 12a. These data inputs are indicated at 36, 37 and 38 respectively. The rate of flow may be measured using a coriolis mass flow meter, so it represents the mass flow rate. The gas composition data enables the calorific value of the gas or gas mixture to be deduced, and in this controller 35 the data signal 37 indicates that calorific value. In the case of the temperature data input 38, this is preferably derived from several temperature sensors such as the thermocouples 39 (only one is shown in reactor block 12a, but three are shown in reactor block 12b of figure 1 ), typically at least ten, distributed through the reactor block 12a, the resulting temperature measurements being subjected to a voting process 40. The voting process 40 includes a device status check which ensures each thermocouple 39 is functional. This check occurs every time the voting process 40 occurs, to ensure the on-going functionality of the thermocouples 39. The status check results in the exclusion of any values which are very different from the others, as indicative of faulty thermocouples; and the remaining values are used to provide the data input 38, which comprises the maximum differential temperature, i.e. the difference between the highest and lowest values obtained; the maximum temperature; and the maximum rate of change of temperature.

The temperature measurements from all the thermocouples 39 are monitored as part of the voting process 40 described above, to check for failure of any of the

thermocouples 39. If one of the thermocouples 39 is identified as having failed, the data from that thermocouple 39 will be excluded from the voting process described above, so that it will utilise only the data from the other thermocouples 39. An invalid measurement from a thermocouple 39 may be identified for example if it is outside a predetermined range of values, or if it differs by more than a predetermined limit from the mean value from other thermocouples 39 in the vicinity, or, in the case of a hardware failure, if the thermocouple 39 fails to give a reading. Where temperatures are measured using thermocouples 39, the thermocouples may be arranged within flow channels, in particular within the combustion channels 16. Alternatively the thermocouples 39 may be located in holes within the sheets that define the walls of the channels, in particular the walls of the combustion channels 16, the holes extending generally within the plane of the sheet so the thermocouples do not directly contact the gases in the reactor block 12.

The observed gas flow rate data 36 is supplied to a control loop feedback controller 46, to which is also provided a flow set point 44. The control loop feedback controller 46 provides an output signal 48 corresponding to any difference between the observed gas flow rate data 36 and the flow set point 44, which is used to control the inlet flow controller 31 .

The observed gas flow rate data 36 is also supplied to a ramp unit 42 to which is also supplied a signal representing the flow set point 44 which may be adjusted in accordance with the observed maximum temperature 38 and rate of change 58 (as described below). The ramp unit 42 is also provided with the observed maximum temperature data 71 , and a target temperature set point 52. The ramp unit 42 provides a ramp signal 45 indicating a currently-desired temperature; during start-up the currently- desired temperature gradually increases for example initially at 1 °C/min, but the rate of increase gradually decreases to zero as the temperature reaches the optimum reactor operating temperature (as described above in relation to figure 5). This ramp signal 45 is provided as an input to a control loop feedback controller 50, for which it provides a set point. The control loop feedback controller 50 is also provided with the observed flow data 36, the gas composition data 37, and the current maximum temperature 71 . The gas composition data 37 in this case indicates the calorific value of the gas, and provides a scaling factor for the ramp signal 45. The control loop feedback controller 50 compares the scaled ramp signal to the observed maximum temperature data 71 , and provides a control signal 54 to adjust the flow set point 44, which as explained above will adjust the inlet flow controller 31 .

The ramp unit 42 is also provided with a control input terminal 43 whereby its operation may be modified, as described below. In an alternative, the gas composition data 37 might instead be supplied to the ramp unit 42. In this case the gas composition data may comprise data on the

components of the gas mixture, from which the ramp unit 42 can calculate the calorific value. In this alternative, the ramp signal 45 provided by the ramp unit 42 would indicate the currently-desired flow rate; and the control loop feedback controller 50 would only have the two inputs, the observed flow rate data 36 and the ramp signal 45.

During start-up, the flow set point 44 is therefore held initially at the current value and then gradually increased to achieve the pre-set temperature set point 52. However if the observed maximum temperature 71 exceeds the temperature set point 52, then the flow set point 44 would be decreased, or vice versa.

Each control loop feedback controller 46, 50 may be a PID controller (Proportional, Integral, Derivative controller), as is widely used in industrial control mechanisms, or may be a more complex electronic controller such as an internal model controller or any form of advanced process controller. It might comprise a microprocessor.

The observed temperature data 38 is also supplied to a rate-of-change unit 56 which provides an output signal 58 corresponding to the maximum rate of change of observed temperature with time. The signal 58 is provided as an input to two comparators 60 and 62. The first comparator 60 is also provided with a first rate-of-change threshold 61 , while the second comparator 62 is also provided with a second rate-of-change threshold 63, the second rate-of-change threshold 63 being larger than the first rate-of- change threshold 61 , and taking into account duration. For example the threshold 61 may be 0.5 °C/s for a period of at least 5 seconds, while the threshold 63 may be 2 ^q C/s occurring for at least 2 seconds. These threshold values are significantly greater than the preferred rate of temperature increase, which in this example is no greater than 1 'O/min, but are sufficiently high that short-term temperature fluctuations are unlikely to trigger the comparators 60 or 62, and are nevertheless significantly less than the rates of change that would occur in a thermal runaway. The values of the thresholds 61 and 63 given here are by way of example only; and it will also be appreciated that the values of the thresholds 61 and 63 may be modified automatically or manually once steady-state operation has been achieved. There may be more than two such comparators 60, 62, all with different thresholds.

If the output signal 58 exceeds the first rate-of-change threshold 61 for the pre-set time limit, a control signal as indicated at box 66 is sent to the terminal 43 of the ramp unit 42, and as indicated at box 74 a signal is sent to reduce the value of the flow set point 44. The signal 66 causes the ramp unit 42 to pause the ramp signal 45 preventing any further increase of the ramp signal 45 during a time delay. This time delay may be pre-set for example at 5 minutes or 10 minutes, or alternatively may be until the observed

temperature data 38 has stabilised. Similarly, if the output signal also exceeds the second rate-of-change threshold 63, then a control signal as indicated at box 66 is sent to the terminal 43 of the ramp unit 42 and as indicated at box 75 a signal is sent to reduce the value of the flow set point 44 by a larger amount. The signal 66 causes the ramp unit 42 to decrease the ramp signal 45 by a predetermined amount, and to impose a pause on any further increase of the ramp signal 45 during the time delay. The decrease of the ramp signal 45 as explained above lowers the effective set point for the control loop feedback controller 46, so that the flow controller 31 reduces the gas flow rate by a larger amount.

It will also be appreciated that there may be a different number of comparators 60, 62, typically up to ten, each provided with a respective rate-of-change threshold 61 , 63. For example the lowest threshold 61 may be set at a value between Ο.Οδ'Ό/ε and 1 °C/s, while the highest threshold 63 may be set at a value between 1 °C/s to 10°C/s. If any threshold is exceeded, a corresponding control signal 74, 75 decreases the flow set point 44 by a corresponding predetermined amount.

In addition the controller 350 includes a comparator 70 to which are provided signals representing the maximum temperature 71 , and a maximum temperature threshold 76. In a similar way to that described above, if the maximum temperature 71 exceeds the maximum temperature threshold 76, for a pre-set period, then a signal 66 causes the ramp 42 to pause, while a signal 77 reduces the flow set point 44.

It will be appreciated that the controller 350 of the second reactor block 12b is substantially the same as that of the first reactor block 12a, although additional data inputs are required. The controller 350 of the second reactor block 12b controls the flow control 33 through which additional fuel is provided. In this case the controller 350 requires data not only on the reactor temperature, but also the flow rate of the air supplied through the inlet 24, the flow rate of the exhaust gases in the duct 22, and the flow rate of the additional fuel supplied through the inlet 26. Furthermore the controller 350 requires the gas composition of the three gas streams: the exhaust gases in the duct 22, the air at inlet 24, and the fuel gas at inlet 26. The controller 350 may also be provided with data on the oxygen concentration remaining in the exhaust gases from the combustion channels 16 of the second reactor block 12b. Preferably all this information on the gas flows and the gas compositions is provided to the ramp unit 42, which can consequently deduce the thermal energy available from combustion of the gas mixture. Hence the ramp unit 42 can produce a ramp signal 45 which indicates the required flow rate of the fuel gas through the inlet 26, and so this ramp signal 45 is the set point for the control loop feedback controller 50. The control loop feedback controller 50 therefore only requires input data 36 representing the observed flow rate of the fuel gas through the inlet 26, in addition to the ramp signal 45 (in addition to data representing constants).

In other respects the control system 350 for the second reactor block 12b operates in the same way as described above. However, it will be appreciated that the controller 350 must know the current status of the first reactor block 12a in order to synchronise start-up of the second reactor block 12b.

Hence, when the reaction module 10 is to be brought into operation, an operator may set the threshold values 61 and 63, and the temperature set point 52, which may for example be set to hold the maximum temperature at 760 °C, in the controllers 35 of both the reactor blocks 12a and 12b. As previously mentioned the reactor blocks 12a and 12b may then be preheated to 550 °C by flowing electrically-heated nitrogen through the flow channels 15 and 16, or by performing combustion for example of an oxygenate such as methanol in the combustion flow channels 16. The steam/methane mixture, which may be preheated to 620 °C, is then gradually supplied to the channels 15.

The methane/air mixture is then supplied through the inlet flow controller 31 to the first reactor block 12a, and air and methane are supplied through the inlet flow controllers 32 and 33 to the inlets 24 and 26. The flows through the inlet flow controller 31 and through the inlet flow controller 33 are controlled by the controllers 35 or 350 as described above.

Each controller 35 or 350 may be described as having three different modes of operation: a ramp mode; an anti-runaway mode; and a ramp-up after cutback mode. These are summarised in the following three paragraphs.

Referring first to the controller 35 of figure 4, as explained above, the control loop feedback controller 50 in each case provides a value for the flow set point 44, this being the expected steady-state value for the flow rate of fuel through the inlet flow controllers 31 and 33 respectively. The observed flow data 36 will initially be considerably less than the flow set point 44, so the ramp signal 45 from the ramp unit 42 instructs the control loop feedback controller 46 to gradually increase the fuel flow rate. The variation of the ramp signal 45 with time therefore controls the rate at which the fuel flow rate and the temperature are brought up to their desired, steady-state values. If the observed average temperature 38 exceeds the temperature set point 52, the control loop feedback controller 50 decreases the flow set point 44, and consequently the ramp unit 42 decreases the ramp signal 45, so the control loop feedback controller 46 decreases the fuel flow. Hence the reactor blocks 12a and 12b are brought up to the pre-set operating temperature, and are then held at that temperature. This mode of operation is the ramp mode.

Referring now to the controller 350 of figure 7, the control loop feedback controller 50 provides a value for the flow set point 44, this being the required value for the flow rate of fuel through the inlet flow controller 31 or 33. The observed flow data 36 will be controlled by the control loop feedback controller 50 to flow set point 44, so the ramp signal 45 from the ramp unit 42 instructs the control loop feedback controller 50 to gradually increase the fuel flow rate. If the observed maximum temperature 71 exceeds the temperature set point 52, the control loop feedback controller 50 decreases the flow set point 44, so the control loop feedback controller 46 decreases the fuel flow. Hence the reactor blocks 12a and 12b are brought up to the pre-set operating temperature, and are then held at that temperature. This mode of operation is the ramp mode. It may be applied to a ramp-up or to a ramp-down.

In a further modification, if the temperature 38 exceeds the temperature set point 52, the temperature first enters a deadband, which is a temperature band of

approximately 1 -5°C. Once the temperature 38 exceeds the temperature set point 52 by the width of the deadband, a pause signal 66 is generated, to cause the ramp 42 to pause until the temperature 38 falls within the deadband. In order to achieve this, the ramp rate and constants may be adjusted. The pause allows for the reactor conditions to stabilise in order to avoid over-correction.

If at any point a thermal runaway occurs, this being most likely during the final stages of the approach to steady-state operation, the rapid increase in temperature is detected by one or both of the comparators 60, 62. Consequently the fuel supply is decreased by the signal 64 or 68 (in the case of the controller 35), or by the signal 74 or 75 (in the case of the controller 350), and is held steady for the pre-set time delay by the pause signal 66. As mentioned above, there may also be a high maximum temperature threshold 76, and a comparator 70, to instigate the same cutback. This mode of operation is the anti-runaway mode.

If the temperature of the reactor block 12a or 12b stabilises below the pre-set operating temperature, then after that time delay the ramp unit 42 will again provide a ramp signal 45 to instruct the control loop feedback controller 46 to increase the fuel flow again. This mode of operation is the ramp-up after cutback mode. Hence the controllers 35 and 350 ensure that the reactor blocks 12a and 12b are heated up to the respective desired operating temperatures in a controlled fashion, without risking thermal runaway. The ramp mode may be also utilised when the operation of the reactor block

12 is to be closed down, so in this case the fuel rate would be gradually decreased in accordance with a decreasing ramp signal 45, so the temperature decreases at a desired rate. It will also be appreciated that the controllers 35 and 350 may be modified, without departing from the scope of the present invention. In particular, as described above, a single signal representing all the observed temperature data 38 is provided to the controller 35. In a modification the temperature and its rate of change as measured by each thermocouple 39 may be monitored separately. If any one of those measured temperatures exceeds the predetermined threshold, then the fuel supply would be decreased by the signal 64 or 68 as described above. Furthermore, as in the controller 350, the maximum value 71 of temperature within all the combustion channels 16 may be monitored, and if this exceeds a preset threshold 76 the controller 35 may be arranged to reduce the fuel flow rate.