Process for preparing alkylene oxide addition products

Abstract

A process for preparing alkylene oxide addition products, comprises (i) purging a tank reactor with an inert gas to displace any oxygen; (ii) charging the tank reactor with a starting compound capable of adding on or inserting an alkylene oxide; (iii) feeding an alkylene oxide into the tank reactor; (iv) determining: a flow- limited feeding rate r1, as a function of the total amount of alkylene oxide that is already fed; a cooling rate-limited feeding rate r2, as a function of the utilization of the cooling capacity; and an unreacted alkylene oxide accumulation-limited feeding rate r3, as a function of whether a runaway pressure approaches or exceeds the maximum allowed operation pressure of the reactor and / or a runaway temperature approaches or exceeds the maximum allowed operation temperature of the reactor or the onset temperature of a secondary decomposition reaction, whichever is lower; (v) determining the minimum rmin of r1, r2 and r3; and (vi) controlling the feeding rate of the alkylene oxide to rmin.

Description

[0001] Process for Preparing Alkylene Oxide Addition Products

[0002] Addition products of alkylene oxides onto alcohols, amines, acids or esters are important industrial products which find a variety of uses, especially as nonionic surfactants.

[0003] The alkoxylation is typically performed semi-batchwise, for example in stirred autoclaves or loop reactors, at temperatures between 80 and 200 °C. Alternatively, the liquid reaction mixture can also be dispersed into an alkylene oxide-containing gas phase. Typically, a compound with the nucleophilic site - for example an alcohol, a carboxylic acid, an ester or an amine - is initially charged together with the catalyst and then the desired amount of alkylene oxide is dosed thereto, which generally establishes a pressure of up to 20 bar depending on the temperature. Suitable catalysts are basic compounds, for example alkali metal alkoxides, or Lewis acids.

[0004] For safety reasons owing to the flammability of alkylene oxide, the process is typically performed under inert atmosphere, such as nitrogen or another inert gas. The partial pressure of the alkylene oxide reactant is preferably limited by dilution with the inert gas, to a gas phase concentration which is below the lower decomposition limit of the alkylene oxide. This ensures safety against explosion-like polyaddition of highly reactive alkylene oxides, such as ethylene oxide.

[0005] Likewise, the rate of addition of the alkylene oxide is limited such that even in case of a runaway reaction, the reactor pressure and temperature are still within the design limits of the reactor and below the onset temperature of a secondary decomposition reaction. Without knowing the progress of the reaction within the reactor, this means that large safety margins must be adopted.

[0006] Dilution of the alkylene oxide reactant with an inert gas and slow addition rates, however, reduce the alkylene oxide equilibrium concentration of the liquid phase due to lower alkylene partial pressure in the gas phase. Reaction rate is slowed because liquid alkylene oxide concentration is low.

[0007] US 2009 / 326283 discloses a method for the production of alkylene oxide addition products comprising (I) charging a stirred reactor with a starting compound plus a diluent gas, wherein a portion of said alkylene oxide reacts in a liquid phase with said starting compound, and the remaining alkylene oxide together with said diluent gas forms a gas phase, (II) continuously drawing off said liquid phase from the bottom of the reactor, and recycling to the top of the reactor via a Venturi nozzle, and (ill) metering said gas phase comprising alkylene oxide into the Venturi nozzle via a vacuum line.

[0008] US 2012 / 0035381 discloses a process for preparing polyether alcohols, wherein an alkylene oxide is introduced into a reactor in such a way that the concentration of unreacted alkylene oxide in the liquid reaction mixture in the reactor is increased as far as possible with increasing degree of addition of the alkylene oxide onto the starter, with the proviso that the temperature at the end of a runaway reaction is at any point in time at least 100 K below the onset temperature of the decomposition reaction of the liquid reaction mixture. The present invention relates to methods for controlling alkylene oxide addition reactions. The invention seeks to provide a process for preparing alkylene oxide addition products that can be safely accomplished at greater reaction rates within shorter time frames.

[0009] The invention relates to a process for preparing alkylene oxide addition products, comprising:

[0010] (I) purging a tank reactor with an inert gas to displace any oxygen;

[0011] (ii) charging the tank reactor with a starting compound capable of adding on or inserting an alkylene oxide;

[0012] (ill) feeding an alkylene oxide into the tank reactor;

[0013] (iv) determining:

[0014] - a flow-limited feeding rate r1 , as a function of the total amount of alkylene oxide that is already fed;

[0015] - a cooling rate-limited feeding rate r2, as a function of the utilization of the cooling capacity; and

[0016] - an unreacted alkylene oxide accumulation-limited feeding rate r3, as a function of whether a runaway pressure approaches or exceeds the maximum allowed operation pressure of the reactor and / or a runaway temperature approaches or exceeds the maximum allowed operation temperature of the reactor or the onset temperature of a secondary decomposition reaction, whichever is lower, wherein the runaway temperature and the runaway pressure are the temperature and the pressure which occur in the case of a spontaneous adiabatic runaway reaction of the accumulated amount of unreacted alkylene oxide;

[0017] (v) determining the minimum rmin of r1 , r2 and r3; and

[0018] (vi) controlling the feeding rate of the alkylene oxide to rmin.

[0019] The invention is based on the insight that generally several limitations need to be considered for safeguarding alkylene oxide addition reactions. These limitations are reflected by feeding rates r1 , r2 and r3. The calculated feeding rates are fed into a minimum selection and the lowest feeding rate rmin, determining the current maximum allowed alkylene oxide feeding rate, is fed to a flow controller as a setpoint. The feeding rates r1 , r2 and r3 may be reported as mass flow rates or as a percentage of a maximum flow rate. The smallest percentage value specifies the limit. The minimum percentage is used in a central flow controller to set the flow rate, or can also directly specify the opening degree of the valve.

[0020] A runaway reaction can occur for various reasons. If insufficient cooling or a complete cooling failure occurs during alkylene oxide addition, the heat released by the exothermic reaction of the unreacted alkylene oxide present in the reactor results in an increase in the reactor pressure and temperature. A temperature increase will cause a faster reaction rate, and this in turn will produce more heat and a higher temperature. The resulting runaway reaction is almost completely adiabatic because the rise in temperature and pressure can no longer be limited by cooling.

[0021] While the concentration of unreacted alkylene oxide constitutes a driving force for the addition of alkylene oxide to the growing alkylene oxide chain of the addition product, excessive accumulation of unreacted alkylene oxide must be avoided. Spontaneous polyaddition reaction of the unreacted alkylene oxide may cause a runaway to occur which cannot be controlled anymore by the available cooling capacity. During a runaway reaction, no safety -critical conditions occur as long as the design conditions of the reactor or the onset temperature of any secondary reaction are not reached. If these are exceeded, however, the reactor could be damaged or even destroyed. In a worst case scenario, when sufficiently high temperatures are reached, secondary reactions can occur which reach the proportions of a heat explosion that could rupture the reactor. These situations must be reliably prevented because of the great potential hazard to people, the plant and the environment.

[0022] Flow-Limited Feeding Rate r1

[0023] In the process, the starting materials (principally the starting compound capable of adding on or inserting an alkylene oxide) are introduced in pre-specified metering amounts into a tank reactor to which an alkylene oxide is fed. The flow-limited feeding rate r1 is a function of the total amount of alkylene oxide that is already fed. It constitutes a recipe- and / or reactor-specific feeding rate restriction.

[0024] The relationship between the maximum allowed feeding rate and the total amount of alkylene oxide that is already fed is generally based on experience gathered from previous batches. The flow-limited feeding rate r1 may also accommodate restrictions due to reactor design or reactor periphery, e.g., maximum allowed flow rates and other safety limits. A first set of flow curves can be calculated beforehand by use of simulation tools.

[0025] This relationship can be depicted as an alkylene oxide feeding curve, which is a function of the maximum allowed feeding rate vs. the total amount of alkylene oxide that is already fed from which feeding rate r1 may be retrieved. This flow curve will be implemented as separate coordinates (x,y), where the x-value represents the total amount of alkylene oxide that is already dosed, and the y-value represents the maximum allowed feeding rate.

[0026] In one embodiment, the flow-limited feeding rate r1 is based on a computer memory, the memory having stored therein an alkylene oxide feeding curve as discussed above.

[0027] Feeding rate r1 is envisaged to be the limiting factor particularly during start-up and at early stages of the reaction with low conversions. This is due to the fact that at this stage of the process, only a low amount of reaction enthalpy has been released, hence utilization of the cooling capacity is low. Moreover, the low amount of alkylene oxide does not constitute a hazard even under the assumption of a spontaneous adiabatic runaway reaction. An initial limitation of the maximum alkylene oxide feeding rate prevents too fast dosing in the early stages of the alkylene oxide dosing step, which can lead to instable process conditions due to, e.g., area low heat exchange area because of a low liquid level in the reactor. At the end of the batch, the maximum feeding rate is normally determined by r2 or r3. By using this approach, the feeding rate of alkylene oxide can be maximized while operating within safety limits.

[0028] In one embodiment, feeding rate r1 is determined in accordance with the following equation: rl = al + bl * moxidewherein a1 [kg / h] is a feeding rate, b1 [1 / h] is a gradient, and mOxide is the amount of alkylene oxide that is already fed [kg]. Once the envisaged total amount of alkylene oxide has been fed, feeding rate r1 is reduced to 0 kg / h. The feeding rate a1 is not particularly limited and depends on the dimensions of the envisaged reactor and process. Feeding rate a1 may be in the range of 1 to 20,000 kg / h, such as 10 to 15,000 kg / h, for example 500 to 5,000 kg / h. The gradient b1 may be in the range of 0.0 to 2.0 per hour (IT1), more preferably 0.3 to 0.7 IT1, such as 0.5 IT1.

[0029] The feeding rate r1 may be varied depending on the amount mOxide of alkylene oxide that is already fed. For example, where a total amount of alkylene oxide of 3,000 kg is envisaged, the feeding rate may be ramped up in accordance with the equation above until mOxide corresponds to 1 ,000 kg, and then kept constant. Thus, in a further embodiment, feeding rate r1 is determined in accordance with the following equation for the range of 0 kg < mOxide (y * mOxide_totai): rl = al + bl * moxidewherein y * mOxide_totai [kg] is a fraction of the total amount mOxide_totai of alkylene oxide to be fed to the reactor, with y having a value of 0 < y < 1 , such as 0.1 < y < 0.7, a1 [kg / h] is a feeding rate, b1 [1 / h] is a gradient, and mOxide is the amount of alkylene oxide that is already fed [kg]. The fraction y * mOxide_totai is typically in the range of 10 wt.-% to 70 wt.-%, such as 25 wt.-% to 50 wt.-%, of the total amount mOxide_totai of alkylene oxide to be fed to the reactor. The feeding rate a1 and gradient b1 are as defined above.

[0030] In the range of (y * mOxide_totai) < mOxide mOxide_totai, feeding rate r1 is then determined in accordance with the following equation: rl = cl wherein d [kg / h] is a constant feeding rate. Feeding rate d is not particularly limited, but suitably corresponds to the value of r1 obtained in accordance with the equation above when mOxide = (y * mOxide_totai).

[0031] Once the envisaged total amount of alkylene oxide mOxide_totai has been fed, feeding rate r1 is reduced to 0 kg / h.

[0032] In one embodiment, feeding rate r1 is determined by the analytical equation(s) according to either of the two embodiments above, and the minimum rmin of r1 , r2 and r3 is used as the setpoint of a flow controller.

[0033] In another embodiment, feeding rate r1 constitutes the setpoint of a flow controller, in particular a PID controller. The controller attempts to maximize the flow of alkylene oxide. In this case, setpoint SPricorresponds to r1 in the above equation.

[0034] Cooling Rate-Limited Feeding Rate r2

[0035] Alkoxylation reactions are exothermal reactions, hence heat will be generated during the reaction. Alkoxylation reactors are therefore equipped with cooling equipment such as half pipe jackets, internal coils and / or external heat exchangers. Continuous cooling is necessary to maintain a desired reaction temperature during the alkylene oxide addition process. The cooling capacity is especially relevant for ethylene oxide and propylene oxide additions due to their large reaction enthalpy. At the start of the reaction, heat exchange in reactors without external circulation often suffers from low heat exchange area due to a low liquid level in the reactor. During the reaction, the effective heat exchange area increases, but product viscosity might also build up, which has a negative effect on the heat transfer coefficient. As all of these influencing factors on the heat balance happen simultaneously and are product- and reactor-specific, an automated controller is required to adapt the alkylene oxide feed rate (heat generation) to the actual heat removal capabilities of the reactor. This automatic control ensures that the full cooling capacity of the reactor is always utilized, even taking into account varying climate conditions, e.g., in winter vs. summer. This allows to minimize alkylene oxide feed times in stages where the cooling capacity is not fully utilized.

[0036] When heat generation is higher than cooling capacity, the target reactor temperature cannot be maintained and temperature increases, although coolant flow is maximized and coolant inlet temperature is as low as possible.

[0037] The aim of feeding rate r2 is to reduce the alkylene oxide feeding rate once the full cooling capacity is reached or exceeded to maintain a stable temperature control in the reactor. As long as the cooling capacity is not fully utilized, the feeding rate r2 does not have a limiting effect on the alkylene oxide feed rate.

[0038] The remaining cooling capacity can be determined in various manners, for example by at least one of the following parameters:

[0039] - the coolant inlet flow temperature;

[0040] - the difference between the current reactor temperature and a predetermined reactor temperature;

[0041] - the coolant flow;

[0042] - the opening degree of a coolant flow valve;

[0043] - the difference between the current reactor temperature and the coolant inlet temperature;

[0044] - temperature difference between the coolant inlet flow and the coolant return flow; and

[0045] - heat balancing of the coolant inlet flow and the coolant return flow.

[0046] Also envisaged are logic combinations of two or more of the above parameters, such as the opening degree of a coolant flow valve and the difference between the current reactor temperature and a predetermined reactor temperature; or the coolant flow and the difference between the current reactor temperature and a predetermined reactor temperature.

[0047] In one embodiment, feeding rate r2 is determined in accordance with the following equation: r2 = aR* F * VR(moxide) wherein aR is the specific heat transfer area per liquid reactor volume [m2 / m3], F is a measure of the heat transfer capacity of the rector [kg (al ky lene oxide) I (m2* s)], and V (mOxide) is the volume of the liquid reaction component [m3]. This equation is applicable under the assumption that the specific cooling surface inside the reactor is approximately constant and the heat transfer resistances do not change significantly. For example, F may be calculated as follows: wherein TR is the desired reactor temperature [°C], TCOoi,min is the minimum temperature of the cooling means [°C], a is the heat transfer coefficient [kW I m2K] and AHR. AO is the specific heat of reaction [kJ I kg].

[0048] In particular, feeding rate r2 may be determined by the analytical equation above, and the minimum rmin of r1 , r2 and r3 is used as the setpoint of a flow controller.

[0049] In another embodiment, feeding rate r2 constitutes the setpoint of a flow controller, in particular a PID controller. The controller attempts to exceed the setpoint of the internal reactor temperature SPTR by a permitted amount AT, such as 3 K. In one embodiment, the setpoint SPr2, i.e., feeding rate r2 at a particular point of time during the process, is calculated as follows:

[0050] SPr2=PTR "F T

[0051] If the cooling limitation is active, i.e., r2 = rmin, the offset of r1 and r2 ensures that both the cooling control is maximized (full opening of the cooling valve) and the oxide flow is so high that the internal reactor temperature is maximized within the permitted operating window. The oxide flow then effectively regulates the internal reactor temperature, while the cooling valve is fully opened.

[0052] Unreacted Alkylene Oxide Accumulation-Limited Feeding Rate r3

[0053] The aim of feeding rate r3 is to identify potential states which could result in the admissible operating pressure or temperature being exceeded in good time before a risk can arise. The unreacted alkylene oxide accumulation-limited feeding rate r3 is a function of whether a runaway pressure approaches or exceeds the maximum allowed operation pressure of the reactor and / or a runaway temperature approaches or exceeds the maximum allowed operation temperature of the reactor or the onset temperature of a secondary decomposition reaction, whichever is lower. The runaway pressure and runaway temperature are the temperature and the pressure which occur in the case of a spontaneous adiabatic runaway reaction of the accumulated amount of unreacted alkylene oxide.

[0054] In one embodiment, the maximum allowed accumulated amount of unreacted alkylene oxide throughout the process can be calculated beforehand by setting certain boundary conditions. The boundary conditions include a minimum reaction temperature and minimum starter amounts, and may include the inertization conditions of the tank reactor. A control parameter that correlates with the amount of unreacted alkylene oxide is measured during the reaction. A preferred control parameter is the current reactor pressure. The vapor pressure of the alkylene oxide contributes to the total pressure in the tank reactor (provided that the minimum temperature is sufficiently high). Hence, the current reactor pressure is considered a representative indicator for alkylene oxide accumulation.

[0055] Thus, determining the unreacted alkylene oxide accumulation-limited feeding rate r3 may comprise establishing a pre-determined shutoff pressure and comparing the current reactor pressure to the pre- determined shutoff pressure. When the current reactor pressure approaches or exceeds the pre-determined shutoff pressure, r3 is reduced or set to zero.

[0056] The pre-determined shutoff pressure may be a fixed shutoff pressure, which is independent of the reaction progress. However, the pre-determined shutoff pressure is preferably a function of the total amount of alkylene oxide that is already fed. As alkylene oxide is fed into the reactor, the volume of the liquid phase increases, compressing the gas phase and causing the inert gas partial pressure to rise. With the same allowable partial pressure of unreacted alkylene oxide, the reactor pressure is higher. Hence, the predetermined shutoff pressure will increase as the total amount of alkylene oxide that is already fed increases.

[0057] Comparison of the calculated runaway pressure and temperature within the design limits for the reactor provides information on the safety reserves still present. These safety reserves can be used to optimize operation, for example to increase the feed rate or raise the reaction temperature. The runaway pressure and temperature is calculated continuously throughout the reaction, enabling measures for safe termination of the reaction to be taken in good time, in particular before an actual runaway can be measured at all.

[0058] Feeding rate r3 relies on determining the concentration of unreacted alkylene oxide in the reactor during the course of the reaction. The amount of unreacted alkylene oxide inside the reactor is the key quantity in modelling al koxy lation reactions. The amount of unreacted alkylene oxide is related to the reaction rate and is thus of relevance with respect to process optimization. At the same time, it is the crucial quantity and to be limited in terms of process safety.

[0059] The alkylene oxide concentration may be determined directly or indirectly.

[0060] In one embodiment, the amount of unreacted alkylene oxide within the reactor is determined via online spectrometry or online calorimetry.

[0061] An indirect method of determining the unreacted alkylene oxide concentration which is simple to use and can be employed universally has been found to be measurement of the pressure in the reactor. If the alkylene oxide is dosed faster than it can react, accumulated unreacted oxide leads to an increase in the partial pressure of the alkylene oxide, which results in a rise of overall reactor pressure. Hence, the concentration of unreacted alkylene oxide can be calculated when the solubility coefficient of the alkylene oxide in the reaction mixture is known and the compression pressure of the inert gas, such as nitrogen, has been calculated. As alkylene oxide is fed into the reactor, the volume of the liquid phase increases, compressing the gas phase and causing the inert gas partial pressure to rise. As a result, some of the inert gas in the gas phase dissolves into the liquid phase.

[0062] Hence, in an embodiment, the amount of unreacted alkylene oxide within the reactor is determined via a thermodynamic model on the basis of

[0063] - the total amount of alkylene oxide that is already fed,

[0064] - the current reactor pressure and the current reactor temperature, and

[0065] - the amount of starting compound. With knowledge of the amount of accumulated unreacted oxide, the temperature and the pressure which occur in the case of a spontaneous adiabatic runaway reaction can be calculated based on the thermodynamic model. The calculated runaway temperature and runaway pressure can then be compared to the design limits of the reactor, and / or to the onset temperature of a secondary decomposition reaction. The design limits of the reactor regarding temperature and pressure are understood to relate to the maximum allowed operation temperature and the maximum allowed operation pressure of the tank reactor. The maximum allowed operation temperature and the maximum allowed operation pressure are understood to be the maximum temperature and maximum pressure that can be encountered without harm to the tank reactor.

[0066] In some cases, the onset temperature of a secondary decomposition reaction of the liquid mixture in the tank reactor at a given point may be lower than the maximum allowed operation temperature of the vessel. The consequence of exceeding this onset temperature is an exothermic decomposition reaction which cannot be stopped by the remaining cooling capacity of the vessel in case of a loss of energy. In this case, the temperature after the adiabatic runaway of the alkoxylation reaction is limited to below this onset temperature.

[0067] Secondary decomposition reactions occur because the starting compound or alkoxylated products thereof may undergo highly exothermic decomposition reactions when the onset temperature is reached. The onset temperatures of the decomposition reaction of alkoxylated products depends on the degree of addition of the alkylene oxide onto the starting compound.

[0068] Similarly, the resulting pressure during or after the runaway reaction may be limited below the maximum allowed operation pressure of the tank reactor or a safety valve safeguarding the tank reactor, when an adequate pressure relief during and after the runaway is not possible. This may occur due to either too large gas volumes developing during the runaway to be vented through a reasonably sized safety valve or due to the risk of emission of toxic or other harmful substances from the reactor when relieving pressure.

[0069] Within the thermodynamic model, the amount of unreacted, accumulated alkylene oxide nA, defines the energetic potential which may be released in a runaway scenario. During that runaway, the process temperature increases with oxide conversion and can be described as wherein

[0070] Ah / ? denotes the reaction enthalpy cpdenotes the heat capacity of the reactor contents degree of alkoxylation 9 where 9 also is a function of the conversion , 9 = 0( ). Due to the high reactivity the runaway happens on short timescales and is modelled as adiabatic. The maximum temperature occurs at the end of the runaway (full conversion of free oxide, = 1), hence y _ 'T'f -| \

[0071] ‘ max ~1)

[0072] Not only the temperature increases during a runaway, but also the pressure. In the early phase of a runaway, pressure increases due to the temperature increase. With increasing consumption of the free alkylene oxide during the runaway (conversion), pressure decreases again. Thus, the maximum pressure does not occur at the end of the runaway but during the runaway at conversion f+, which can be calculated as

[0073] Pmax=P ”+)

[0074] In order to limit the computational workload, the runaway temperature and pressure may be calculated by a simulated stepwise reaction of the alkylene oxide present in the reactor in k steps. A temperature increase AT arising in the reactor after each step arises from the amount of heat formed (reaction enthalpy), giving rise to a new intermediate temperature. The amounts of alkylene oxide remaining in the reactor and products formed by the reaction are re-calculated after each virtual reaction of a partial quantity An.

[0075] A suitable equation for determining feeding rate r3 should describe the safe dosage rate of alkylene oxide over the course of the reaction. This can be reliably determined experimentally using reaction calorimetry and scaled to the conditions in production. The equation for r3 hence may be obtained from experiment data. For a typical system, a constant accumulation of alkylene oxide is acceptable. Therefore, the permitted dosing rate increases proportionally with increasing reaction volume. This results in a linear equation. In one embodiment, feeding rate r3 may be determined in accordance with the following equation:

[0076] ^3 Real * moxide d" C wherein Real is the gradient of the line (corresponding to the specific increase in the dosing rate over the dosed quantity of alkylene oxide), mOxide is the amount of alkylene oxide that is dosed, and C is the permitted dosing rate of alkylene oxide to the starting compound.

[0077] In particular, feeding rate r3 may be determined by the analytical equation above, and the minimum rmin of r1 , r2 and r3 is used as the setpoint of a flow controller.

[0078] In another embodiment, feeding rate r3 constitutes the setpoint of a flow controller, in particular a PID controller. The controller regulates the oxide flow in the direction of a specific safety limit. This is specified by the maximum concentration of unreacted alkylene oxide. This concentration can be determined indirectly by a process variable, for example the reactor pressure, or calculated via a heat balance of the reaction system.

[0079] The permitted alkylene oxide concentration (known from the safety-related data of the system) can be converted into a partial pressure via the gas solubility at the reactor temperature, for example, which is added to the compression behavior of the inert gas. This results in a curve that can often be described with a simple equation, for example, derived from the ideal gas law.

[0080] The compression of the inert gas phase may be described by

[0081] Pi, O^o Pi m-oxide ^ -oxide with

[0082] . . _ , ,0xide moxLde 0 PL wherein Vo is the volume of the gas phase at the start of reaction [m3], Vm(oxide) is the volume of the gas phase when a defined amount of alkylene oxide has been fed [m3], PL is the density of the liquid phase [kg / m3], and mOxide is the amount of alkylene oxide that is already fed [kg], pi.m(oxide) is the pressure in the compressed inert gas phase, and pi.o is the gas phase pressure at the start of the reaction.

[0083] To this functional term, which describes the compression in the reactor, may now be added an oxide partial pressure which takes into account safety and process technology (PAO). Often, this may be described by a constant value. This yields setpoint 3 (SPr3), i .e. , feeding rate r3: with PAO being the allowed partial pressure of alkylene oxide [bar], e.g., a constant. The setpoint of the third controller can thus be described as a function of the start pressure and reaction progress.

[0084] In one embodiment, each of r1 , r2 and r3 is determined in accordance with the analytical equations described herein, and the minimum rmin of r1 , r2 and r3 is used as the setpoint of a flow controller.

[0085] In another embodiment, each of r1 , r2 and r3 constitutes a set point of a flow controller, in particular a PID controller. The minimum output of the three controllers is selected to determine rmin.

[0086] In a preferred embodiment, the process comprises

[0087] - acquiring process data including reactor temperature, reactor pressure, the total amount of starting compound before the addition of alkylene oxide and the total amount of alkylene oxide that is already fed; - predicting, on the basis of an embedded model and the acquired process data, a future response of the process to changes in the feeding rate of the alkylene oxide; and

[0088] - taking the predicted future response into account in the determination of r3.

[0089] The acquired process data may further include a measured variable indicative of the utilization of cooling capacity.

[0090] The acquired process data may be current process data, data acquired from previous batches or assumed typical process conditions.

[0091] Suitably, the feeding rate of the alkylene oxide is controlled to rmin via a distributed control system (DCS) comprising a controller, the controller being selected from a proportional-integral-derivative (PID) controller and a model predictive controller, in particular a nonlinear model predictive controller, or a combination thereof.

[0092] A model predictive controller must contain appropriate models so as to allow for predictions regarding the behavior of the reaction system and the reactor as well as its peripherals. The following systems should be modelled:

[0093] - reaction system (kinetics and thermodynamics of the reaction);

[0094] - heat balance of reactor and periphery (amount of heat that can be dissipated and dynamic behaviour of the system);

[0095] - safety system (calculation of the safe limits of the operating state);

[0096] - alkylene oxide supply (consideration of maximum flow rates, recipe quantities and possible dependencies with other consumers).

[0097] The distributed control system includes a host computer coupled to one or more controllers over a communication link. The host computers enable end-users to configure, monitor, initiate, and terminate control operations of a controlled process through interaction with the controllers. The controller executes control applications and provides outputs to actuate the valve controlling the feeding rate of the alkylene oxide.

[0098] The process according to the invention comprises the steps of

[0099] (I) purging a tank reactor with an inert gas to displace any oxygen;

[0100] (II) charging the tank reactor with a starting compound capable of adding on or inserting an alkylene oxide,

[0101] (ill) feeding an alkylene oxide into the tank reactor.

[0102] Generally, thorough mixing of the contents of the reactor should be ensured in all the reaction phases. To this end, an agitator, an external circulation system or a combination of both may be adopted. The reactor preferably comprises an agitator which is preferably freely suspended in the reactor. As a further means to increase the mixing intensity, a combination of a tank reactor with an external circulation system may be adopted. Via the external circulation system, liquid reaction mixture is withdrawn from the bottom of the reactor and reintroduced at the top of the reactor or below the liquid level. The external circulation system generally comprises a pump, and preferably may contain at least one heat exchanger.

[0103] The external circulation loop is suitably put into operation once a minimum liquid filling level has been attained. To this end, a discharge valve at the bottom of the reactor is opened and the liquid phase is pumped in circulation and introduced again at the top of the reactor.

[0104] The alkylene oxide addition reaction is highly exothermic. The temperature within the reactor is kept at the desired level or adjusted to the desired level by cooling. Cooling is generally carried out via the reactor wall and / or by means of heat exchanger surfaces arranged internally in the reactor and / or externally in the pumped circulation, e.g. on cooling coils, cooling cartridges, plate, tube bundle or mixer heat exchangers. These should be configured such that cooling can also be carried out effectively at the start of the metering phase, i.e., at a low level of fill. Internal cooling means, especially cooling coils, are suitably located in close proximity to the reactor wall, in particular so as not to interfere with the movement of the agitator, if present

[0105] Suitable coolants for passing through the cooling means are not particularly limited and include water, mixtures of water and glycol, and / or oil. Preferably, the coolant is water.

[0106] The heat exchanger means may also be used at the start of the reaction for heating the contents of the reactor to the reaction temperature.

[0107] The tank reactor is inertized, i.e., purged with an inert gas to displace any oxygen and other gases that may be detrimental to the addition reaction. Preferably, the tank reactor is purged with an inert gas and evacuated several times. This is done by means known to the skilled person for this purpose, such as vacuum pumps or the like. Alternating application of reduced pressure and purging with inert gas removes air and traces of water. Nitrogen, argon or carbon dioxide may be used as inert gases. For reasons of its better availability and thus of its low price, nitrogen is the preferred inert gas. Preferably, oxygen is displaced to below 3 vol.- %, in particular to below 0.3 vol.-%.

[0108] A starting compound capable of adding on or inserting alkylene oxide is charged to the reactor, suitably together with a catalyst, to form a liquid phase.

[0109] Typical examples of starting compounds capable of adding on or inserting alkylene oxides are alcohols, acids such as carboxylic acids, esters such as alkyl carboxylates or polyol carboxylates, amines, hydrogen sulfide and thiols.

[0110] Preferred alcohols are compounds of the formula (I)

[0111] R1-OH (I) wherein R1is a linear or branched hydrocarbon radical having 1 to 22 carbon atoms, preferably 8 to 18 carbon atoms, and 0 or 1 to 3 double bonds. Typical examples are, in addition to the lower aliphatic alcohols methanol, ethanol and the isomeric butanols and pentanols, the fatty alcohols, specifically caproic alcohol, capryl alcohol, 2 -ethylhexyl alcohol, capric alcohol, lauryl alcohol, isotridecyl alcohol, myristyl alcohol, cetyl alcohol, palmoleyl alcohol, stearyl alcohol, isostearyl alcohol, oleyl alcohol, elaidyl alcohol, petroselinyl alcohol, linolyl alcohol, linolenyl alcohol, elaeostearyl alcohol, arachyl alcohol, gadoleyl alcohol, behenyl alcohol, erucyl alcohol and brassidyl alcohol, and technical-grade mixtures thereof, which are obtained, for example, in the high-pressure hydrogenation of technical-grade methyl esters based on fats and oils, or aldehydes from the Roelen oxo process, and as a monomer fraction in the dimerization of unsaturated fatty alcohols.

[0112] Technical-grade fatty alcohols having 12 to 18 carbon atoms, for example coconut fatty alcohol, palm fatty alcohol, palm kernel fatty alcohol or tallow fatty alcohol may be used.

[0113] Preferred acids are carboxylic acids of the formula (II)

[0114] R2-COOH (II) wherein R2is a linear or branched acyl radical having 1 to 22 carbon atoms and 0 or 1 to 3 double bonds. Typical examples are in particular the fatty acids, specifically caproic acid, caprylic acid, 2-ethylhexanoic acid, capric acid, lauric acid, isotridecanoic acid, myristic acid, palmitic acid, palmoleic acid, stearic acid, isostearic acid, oleic acid, elaidic acid, petroselic acid, linoleic acid, linolenic acid, elaeostearic acid, arachic acid, gadoleic acid, behenic acid and erucic acid, and technical-grade mixtures thereof, which are obtained, for example, in the pressure cleavage of natural fats and oils, in the reduction of aldehydes from the Roelen oxo process, or the dimerization of unsaturated fatty acids.

[0115] Technical-grade fatty acids having 12 to 18 carbon atoms, for example coconut fatty acid, palm fatty acid, palm kernel fatty acid or tallow fatty acid may be used. It will be appreciated that it is also possible to ethoxylate functionalized carboxylic acids, for example hydroxycarboxylic acids such as ricinoleic acid or citric acid, or dicarboxylic acids such as adipic acid.

[0116] Suitable esters include use esters obtained from carboxylic acids with alcohols having 1 to 22 and preferably 1 to 4 carbon atoms or polyols, especially glycerol, trimethylolpropane or pentaerythritol. When they are full esters, the alkylene oxide groups are inserted into the carbonyl ester bond.

[0117] Suitable amines include compounds of the formula (III)

[0118] R3-NH-R4(III) wherein R3and R4are each independently hydrogen, alkyl groups having 1 to 18 carbon atoms, hydroxyalkyl groups having 1 to 4 carbon atoms or aminoalkyl groups having 1 to 6 carbon atoms. Typical examples are methylamine, dimethylamine, ethylamine, diethylamine, methylethylamine, and the different propyl, butyl, pentyl and fatty amines of analogous structure. Suitable amines moreover include compounds of the formula (IV)

[0119] B

[0120] [H2N-R]n+i-[NHR]m-[NR]n-NH2 (|V)wherein each R is independently C2-C6 linear alkylene or C3-C6 branched alkylene; m is in the range of 0 to 70; n is in the range of 0 to 35; and B represents a continuation of the structure by branching.

[0121] Typical examples of amines of the formula (IV) include (poly)alkylenepolyamines, such as hexamethylenediamine, diethylenetriamine and triethylenetetraamine, as well as polyalkyleneimines, such as polyethyleneimine.

[0122] Suitable thiols include alkyl thiols such as ethanethiol, propanethiol or butanethiol, as well as alkyl polythiols such as ethane-1 ,2-dithiol and propane-1 , 3-dithiol.

[0123] In general, the molar ratio between alkylene oxide and the starting compounds may be 1 : 1 to 200:1 , preferably 1 : 1 to 50:1 and especially 1 :1 to 20: 1.

[0124] If the starting compound is liquid, the reaction may be carried out in the absence of extraneous solvents. However, if extraneous solvents are necessary or desired, suitable extraneous solvents include non-protic solvents such as ethers, which are not susceptible to alkylene oxide addition, as well as protic solvents such as water and lower alcohols including methanol, ethanol, n-propanol, isopropanol and tert-butanol, and diols including ethylene glycol, diethylene glycol, triethylene glycol, propylene glycol, dipropylene glycol and tripropylene glycol.

[0125] The alkylene oxide addition reaction suitably takes place in the presence of catalysts which may be of homogeneous or heterogeneous nature. Suitable homogeneous catalysts include phosphines, such as triarylphosphines like triphenylphosphine; alkali metals such as metallic sodium and metallic potassium; and alkali metal hydroxides and alkali metal alkoxides, especially sodium hydroxide, potassium hydroxide, sodium methoxide, potassium methoxide, sodium tert-butoxide, potassium tert-butoxide, cesium hydroxide, cesium methoxide, and cesium tert-butoxide.

[0126] Suitable heterogeneous catalysts include double metal cyanide (DMC) catalysts and layered double hydroxides (LDH) catalysts, such as hydrotalcite, which are known in the art. DMC and LDH catalysts have a very high activity in alkylene oxide addition reactions and render possible the preparation of, e.g., polyether polyols under optimum conditions at very low catalyst concentrations (100 ppm or less), so that in general it is no longer necessary to separate off the catalyst from the finished product.

[0127] Suitable heterogeneous catalysts moreover include cyanide-free metal salts, such as zinc chloride, zinc bromide, zinc iodide, zinc acetate, zinc acetylacetonate, zinc benzoate, zinc nitrate, iron(ll) sulfate, iron(ll) bromide, iron(ll) chloride, cobalt(ll) chloride, cobalt(ll) thiocyanate, nickel(ll) chloride and nickel(ll) nitrate.

[0128] Mixtures of various metal salts can also be employed. Alkylene oxide may be metered into the reactor through feeding means, in particular nozzles in the bottom of the tank reactor, through a distributor ring immersed in the liquid or into a recirculation stream of the reactor, if present. The feeding means may be liquid feeding means or gas feeding means. Typically, the alkylene oxide is liquid prior to being metered into the reactor and evaporates upon introduction. The introduction of stirring energy into the liquid phase causes mixing, during which a small amount of the alkylene oxide is dissolved in the liquid phase; the remaining amount forms a gas phase above the liquid with the gas used for inertization.

[0129] The process of the invention is typically conducted discontinuously. Once an adequate degree of alkylene oxide addition has been achieved, the reaction product is withdrawn from the reactor.

[0130] The invention is further illustrated by the following working examples.

[0131] Example 1

[0132] A process for preparing alkylene oxide addition products is carried out by purging a tank reactor with an inert gas to displace any oxygen; charging the tank reactor with a starting compound capable of adding on or inserting alkylene oxide; and alkylene oxide into the reactor.

[0133] The reactor has a free volume of 5 m3. A starting compound amount of 1 ,000 kg with a density of 1 ,000 kg / m3is introduced (gas volume of the reactor at the start of the reaction Vo = 4 m3). 3000 kg of alkylene oxide is dosed in, and the density of the liquid system is constant PL=1 ,000 kg / m3. The liquid volume at the end is 4,000 kg 1 1 ,000 kg / m3= 4 m3(the gas volume of the reactor is 1 m3).

[0134] The tank reactor comprises three PID controllers. Feeding rates r1 , r2 and r3 each constitute a setpoint of one of the flow controllers. The minimum rmin of r1 , r2 and r3 is determined amongst the three setpoints.

[0135] PID controller 1 is configured to set the maximum permitted flow rate. The setpoint SPri(feeding rate r1) for the range of 0 kg < mOxide ^ 1 ,000 kg, i.e., the dosage of the first 1 ,000 kg of alkylene oxide, is calculated as follows: kg

[0136] SPrl= 1,000 — + 0.5 n. wherein mOxide is the amount of alkylene oxide that is already fed. After 1 ,000 kg / h of alkylene oxide have been fed, the setpoint SPr1 , i.e. feeding rate r1 , corresponds to 1 ,500 kg / h.

[0137] The setpoint SPri(feeding rate r1) for the range of 1 ,000 kg < mOxide 3,000 kg, i.e., the dosage of the remaining 2,000 kg of alkylene oxide, is set as follows: kg SPrl= 1,500 - - h.

[0138] Thus, the feeding rate r1 for the remaining 2,000 kg of alkylene oxide is kept constant at 1 ,500 kg / h. Once the total amount of 3,000 kg of alkylene oxide has been fed, SPri(feeding rate r1) is reduced to 0 kg / h. PID controller 2 is configured to use the maximum cooling capacity of the reactor. The controller attempts to exceed the setpoint of the internal reactor temperature SPTR by a permitted amount AT, in this case 3 K. The setpoint SPr2, i.e., feeding rate r2 at a particular point of time during the process, is calculated as follows:

[0139] SPr2 — PTR H” T — SP R H” 3 / f

[0140] PID controller 3 regulates the oxide flow in the direction of the safety limit. The permitted alkylene oxide concentration (known from the safety-related data of the system) can be converted into a partial pressure via the gas solubility at the reactor temperature, for example, which is added to the compression behavior of the inert gas. This results in a curve that can often be described with a simple equation, for example, derived from the ideal gas law. The compression of the inert gas phase may be described by

[0141] Pi, 0^0 PLm<:xlde ^m<:xlde with wherein Vo is the volume of the gas phase at the start of reaction [m3], Vm(oxide) is the volume of the gas phase when a defined amount of alkylene oxide has been fed [m3], PL is the density of the liquid phase [kg / m3], and moxide is the amount of alkylene oxide that is already fed [kg], pi.m(oxide) is the pressure in the compressed inert gas phase, and pi.o is the gas phase pressure at the start of the reaction.

[0142] To this functional term, which describes the compression in the reactor, may now be added an oxide partial pressure which takes into account safety and process technology (PAO). Here, PAO may be described by a constant value. This yields setpoint 3 (SPr3), i .e. , feeding rate r3: with PAO being the allowed partial pressure of alkylene oxide [bar], in this case being a constant.

[0143] The minimum rmin of r1 , r2 and r3 is determined by a comparing unit and the feeding rate of the alkyene oxide is controlled to rmin via the PID controllers.

[0144] Example 2

[0145] The same process as described in Example 1 is carried out, except that each of r1 , r2 and r3 is determined in accordance with the following analytical equations, and the minimum rmin of r1 , r2 and r3 is used as the setpoint of an alkylene oxide flow controller. Feeding rate r1 for the range of 0 kg < mOxide 1 ,000 kg, i.e., the dosage of the first 1 ,000 kg of alkylene oxide is determined in accordance with the following equation: kg rl = 1,000 — + 0.5 wherein mOxide is the amount of alkylene oxide that is already fed. After 1 ,000 kg / h of alkylene oxide have been fed, the setpoint SPr1 , i.e. feeding rate r1 , corresponds to 1 ,500 kg / h. The feeding rate r1 for the range of 1 ,000 kg < mOxide 3,000 kg, i.e., the dosage of the remaining 2,000 kg of alkylene oxide, is kept constant at 1 ,500 kg / h. Once the envisaged total amount of alkylene oxide, i.e., 3,000 kg, has been fed, feeding rate r1 is reduced to 0 kg / h.

[0146] Feeding rate r2 is determined in accordance with the following equation: r2 = aR* F * VR(m0Xlde) wherein aR is the specific heat transfer area per liquid reactor volume [m2 / m3], F is a measure of the heat transfer capacity of the rector [kg (alky lene oxide) I (m2* s)], and V (mOxide) is the volume of the liquid reaction component [m3]. This equation is applicable under the assumption that that the specific cooling surface inside the reactor is approximately constant and the heat transfer resistances do not change significantly.

[0147] F is calculated as follows: wherein T is the desired reactor temperature [°C], TCOoi,min is the minimum temperature of the cooling means [°C], a is the heat transfer coefficient [kW I m2K] and HR. AO is the specific heat of reaction [kJ I kg].

[0148] Feeding rate r3 is determined in accordance with the following equation: r3 Real * moxide d" C wherein Real is the gradient of the line (corresponding to the specific increase in the dosing rate over the dosed quantity of alkylene oxide) [1 / h], mOxide is the amount of alkylene oxide that is already fed [kg], and C is the permitted dosing speed to starting compound [kg / h].

[0149] The minimum rmin of r1 , r2 and r3 is determined by a comparing unit and the feeding rate of the alkyene oxide is controlled to rmin via a flow controller, in particular a PID controller.

Claims

Claims1 . A process for preparing alkylene oxide addition products, comprising:(I) purging a tank reactor with an inert gas to displace any oxygen;(ii) charging the tank reactor with a starting compound capable of adding on or inserting an alkylene oxide;(ill) feeding an alkylene oxide into the tank reactor;(iv) determining:- a flow-limited feeding rate r1 , as a function of the total amount of alkylene oxide that is already fed;- a cooling rate-limited feeding rate r2, as a function of the utilization of the cooling capacity; and- an unreacted alkylene oxide accumulation-limited feeding rate r3, as a function of whether a runaway pressure approaches or exceeds the maximum allowed operation pressure of the reactor and / or a runaway temperature approaches or exceeds the maximum allowed operation temperature of the reactor or the onset temperature of a secondary decomposition reaction, whichever is lower, wherein the runaway temperature and the runaway pressure are the temperature and the pressure which occur in the case of a spontaneous adiabatic runaway reaction of the accumulated amount of unreacted alkylene oxide;(v) determining the minimum rmin of r1 , r2 and r3; and(vi) controlling the feeding rate of the alkylene oxide to rmin.

2. The process according to claim 1 , wherein utilization of the cooling capacity is determined by at least one of the following parameters:- the difference between the current reactor temperature and a predetermined reactor temperature;- the position of a coolant valve;- the difference between the current reactor temperature and the coolant inlet temperature; and- balancing the heat capacities of the coolant inlet flow and the coolant return flow.

3. The process according to claim 1 or 2, wherein determining the unreacted alkylene oxide accumulation-limited feeding rate r3 comprises establishing a pre-determined shutoff pressure and comparing the current reactor pressure to the pre-determined shutoff pressure.

4. The process according to claim 3, wherein the pre-determined shutoff pressure is a function of the total amount of alkylene oxide that is already fed.

5. The process according to claim 1 or 2, wherein the accumulated amount of unreacted alkylene oxide within the reactor is continuously determined via online spectrometry or online calorimetry.

6. The process according to claim 1 or 2, wherein the accumulated amount of unreacted alkylene oxide within the reactor is determined via a thermodynamic model on the basis of- the total amount of alkylene oxide that is already fed,- the current reactor pressure and the current reactor temperature, and- the amount of starting compound.

7. The process according to any one of the preceding claims, wherein the flow-limited feeding rate r1 is based on a computer memory, the memory having stored therein an alkylene oxide feeding curve.

8. The process according to any one of the preceding claims, comprising- acquiring process data including reactor temperature, reactor pressure, the total amount of starting compound and the total amount of alkylene oxide that is already fed;- predicting, on the basis of an embedded model and the acquired process data, a future response of the process to changes in the feeding rate of the alkylene oxide; and- taking the predicted future response into account in the determination of r3.

9. The process according to any one of the preceding claims, wherein a) the flow-limited feeding rate r1 is determined in accordance with the following equation: rl = al + bl * moxidewherein a1 [kg / h] is a feeding rate, b1 [1 / h] is a gradient, and mOxide is the amount of alkylene oxide that is already fed [kg]; or b) the flow-limited feeding rate r1 is determined in accordance with the following equation for the range of 0 kg < mOxide (y * mOxide_totai): rl = al + bl * moxidewherein (y * mOxide_totai) [kg] is a fraction of the total amount mOxide_totai of alkylene oxide to be fed to the reactor, a1 [kg / h] is a feeding rate, b1 [1 / h] is a gradient, and mOxide is the amount of alkylene oxide that is already fed [kg], and the flow-limited feeding rate r1 is determined in accordance with the following equation for the Tange of (y moxidejotal) ** moxide — moxidejotal. rl = cl wherein d [kg / h] is a feeding rate, preferably a feeding rate corresponding to the value of r1 obtained in accordance with the equation r1 = a1 + b1 * m0Xide above when m0Xide = (y * moxide. total)-10. The process according to any one of the preceding claims, wherein the cooling rate-limited feeding rate r2 is determined in accordance with the following equation: r2 = aR* F * VR(m0Xide) wherein aR is the specific heat transfer area per liquid reactor volume [m2 / m3], F is a measure of the heat transfer capacity of the rector [kg (alkylene oxide) I (m2* s)], and V (mOxide) is the volume of the liquid reaction component [m3].11 . The process according to any one of the preceding claims, wherein the unreacted alkylene oxide accumulation-limited feeding rate r3 is determined in accordance with the following equation:^3 Real * ^T-oxide + F wherein Real is the gradient of the line (corresponding to the specific increase in the dosing rate over the dosed quantity of alkylene oxide) [1 / h], mOxide is the amount of alkylene oxide that is already fed [kg], and C is the permitted dosing speed to starting compound [kg / h],12. The process according to any one of the preceding claims, wherein the feeding rate of the alkylene oxide is controlled to rmin via a distributed control system (DCS) comprising a controller, the controller being selected from a proportional-integral-derivative (PID) controller and a model predictive controller, in particular a nonlinear model predictive controller, or a combination thereof.

Images