The present invention relates to a method of producing a chemical product from a reaction mixture. In the method a combined transformed regression model is used. This allows for adjustment of chemical or process parameters in response to local climate parameters such as humidity, temperature or air pressure. The invention is also directed towards a system configured for producing a chemical product and a computing platform configured for producing a chemical product.

In the production of polyurethane foams external, climatic factors may lead to variations in the properties of the final product, for example in regards to the raw density and compression deflection. There is often little choice but to dispose of off- specification material if the process is run throughout a longer period of time where the weather conditions may change.

Before the advent of regression models use was made of heuristic rules, based on the experience of shop personnel. Such individual experience may, however, prove to be too inaccurate to account for many simultaneously changing climate factors and their effect on reaction mixture formulations and process parameters.

The publication “Flexible Polyurethane Slabstock Foam: The Influence of Formulation, Climatic Conditions and Storage Conditions on Foam Properties” by R. Schiffauer and C. den Heijer, Journal of Cellular Plastics Jan/Feb 1983, 61 - 64, reports that:

“The following climatic variables were found to have a significant influence on foam properties: atmospheric pressure during production and absolute humidity during production.

The absolute humidity during storage probably also has a significant influence, but owing to the high intercorrelation with absolute humidity during production this effect could not be estimated separately. For compression set, cell count and air flow no significant effects of the climatic variables were found. The effects of the two dominant climatic variables on the remaining foam properties were:

Higher (lower) atmospheric pressures resulted in higher (lower) densities; higher (lower) hardnesses; lower (higher) elongations at break; lower (higher) tensile strengths.

Higher (lower) absolute humidities resulted in no change in densities; lower (higher) hardnesses; higher (lower) elongations at break; higher (lower) tensile strengths. The similarity of the coefficients for the different grades justified the construction of a combined model using both formulation and climatic variables. Thus, the most important foam properties can be calculated by: ln(property) = Co + Ci ln(water) + C2 ln(blow) + C3 ln(TDI index) + C4 In (atmospheric pressure) + C5 ln(absolute humidity)”.

The publication “Mathematical Property Prediction Models for Flexible Polyurethane Foams

- A Comparison between Conventional Slabstock, High Resilience Slabstock and High Resilience Molded Foams” by R. Schiffauer, Journal of Cellular Plastics 1996, 32, pages 318

- 354” reports on a set of foams prepared by varying a range of formulation parameters and that:

“The data were mathematically analyzed by multiple linear regression techniques, using a variety of transformations of the original variables, resulting in equations correlating properties with predicting variables. As a first approach, all data were regressed in their original, linear format. In most cases, however, better fits could be obtained by transformations of either the predicting variables or the resulting properties or both from a linear into a logarithmic format. Further improvements in the resulting equations could be realized by using interaction terms between some predicting properties. When laboratory and production data were analyzed together, a so-called "dummy variable" was assigned (zero for laboratory data and 1 for production results), in order to compute the effect of scaling up on the various properties.”

The publication “Innovative Produktionsoptimierung fur die Blockschaum-Fertigung“ (innovative production optimization for flexible foam production) by Hubert Ehbing, Karl- Heinz Dorner, Hans -Friedrich Walter, Boiko Raffel and R. Nascher in the conference proceedings „PUR 2002 - Automobil - Comfort - Stmktur - Kiihlen - Bauen”, October 2002, pages 192 - 144, discusses a regression model to address the so-called summer/winter effect in the production of flexible foams. The regression model is: ln(y) = a ₀ + a ₁ In ) + a ₂ Info) + a ₃ Info) + å^ ₄ a _jXj with y = compression hardness; xi = NCO index; X2 = water; X3 = absolute humidity; X4 = T65 content; X5 = nozzle pressure; xr, = block height; X7 = belt speed.

The above mentioned publications fail to address 4 points: firstly, there is no data cleansing procedure and preprocessing before applying regression models; secondly, the involvement of higher order regressors in the regression model is lacking or missing; thirdly, the possibility of automatically and symbolically solving for generating from the regression models a formula for each recipe component influenced by the climate condition is lacking or missing and fourthly, the deployment of these formulas in a user interface or embedded in an FPGA of the production plant is lacking or missing.

Without the first two points the influence of climate conditions may have a defective impact on the formulation correction in the production plant. The last two points are beneficial for either giving the plant technician a suggestion to correct the formulation or for performing the correction automatically.

The present invention has the object of at least partially overcoming a drawback in the art. In particular the invention has the object of providing for the production of a chemical product from a reaction mixture where the influence atmospheric parameters on the final product can be compensated by changing the composition of the reaction mixture in a simplified way.

This object is achieved by a method according to claim 1, a system according to claim 13 and a computing platform according to claim 14. Advantageous embodiments are the subject of the dependent claims. They may be combined freely unless the context clearly indicates otherwise. It is also within the scope of the present invention that the subject-matter of embodiments described in connection with the method, particularly the in the dependent method claims, may also be applied to the system and computing platform according to the invention. This extends to combinations of embodiments as well.

Accordingly, the method of producing a chemical product from a reaction mixture containing at least two components comprises:

I) providing a data set comprising a plurality of production instances, each production instance comprising data on (a) the composition of the reaction mixture and/or process conditions, (b) at least one climate parameter at the production site of the product, and (c) at least one target physical property of the product;

II) generating a regression model for each target physical property of the product using the data set, wherein in the regression model a respective target physical property is the dependent variable and at least part of the other data in the data set, excluding any remaining target physical properties, are independent variables;

III) transforming each regression model to obtain a transformed regression model where a) the composition of the reaction mixture and/or b) the process conditions of a reaction depends on at least one climate parameter at the production site of the product and each target physical property is assigned a desired value;

IV) combining the transformed regression models to obtain a combined regression model;

V) measuring, at the production site, the at least one climate parameter used in the combined regression model;

VI) calculating a) the composition of the reaction mixture and/or b) the process conditions for the reaction, using the combined regression model and the at least one measured climate parameter;

VII) composing the reaction mixture and or process conditions according the calculation; and

VIII) reacting the reaction mixture to obtain the chemical product.

The method according to the invention may, at least partially, be a computer-implemented method. Preferably, steps I to VII are computer-implemented steps.

The data set in step I) of the method comprises (represents) a plurality of production instances and may be obtained from data collected as routine process and product monitoring. A production instance may be a single batch or a single campaign in the production of a chemical product. In the case of a continuous production process a production instance may be viewed as a time span in the continuous process. In general, the quality of the data set will improve with greater detail in regards of the production instances.

Besides data on the composition of the reaction mixture and/or the process conditions applicable for the respective production run the production instance comprises data on at least one (preferably two or more) climate parameter at the production site of the product during the respective production run. As this climate data is attributed to the production instance, it may also be described as the respective climate parameters such as air pressure, etc. recorded during the production of the product in the individual instance. In view of the climate parameter(s) recorded, the production of the chemical product is preferably a process where there is at least partially contact of the reaction mixture or one or more of the components of the reaction mixture with the atmosphere.

Lastly, the production instance comprises data on at least one (preferably two or more) target physical property of the product of the respective production run. Such a target physical property in the production instances and therefore in the data set can be viewed as a property which is part of the specifications of the product, which has been determined by measurement for populating the data set and which will be a target to achieve in future production of the chemical product.

In step II), using the data set, a regression model is generated for each target physical property of the product. The target physical property is therefore an output and other data in the data set are inputs. If there are other target physical properties in the data set, they are disregarded. Examples for regression models include least squares sum and maximum likelihood.

In step III) each regression model obtained in step II) is respectively transformed into a model where at least one climate parameter at the production site of the product is an input. It goes without saying that this at least one climate parameter was also included in the data set as described in step I). Furthermore, each target physical property is assigned a value, thereby reflecting the desired outcome of the chemical reaction. It also goes without saying that the target physical property or properties was included in the data set as described in step I). The output of each transformed model is the composition of the reaction mixture and/or the process conditions of a reaction necessary to achieve the target physical properties. In summary, step III) transforms n regression models into n transformed regression models.

Furthermore, the involvement of higher order regressors with interaction can also improve the fit of a regression model. Such a regression model may have the following form:

With the fitted regression models for each property = /(¾)) a symbolic solver can be applied in order to receive formulas for correction the formulation parameters x and x ₂ .

It is advantageous to use data cleansing and preprocessing techniques to enhance the data quality of the historical data and to improve the fit of the regression model(s). The data cleansing techniques are preferably focused to find outliers and impute them either with a default value or with a statistical descriptor such as the mean and the median of the corresponding parameter.

Hence, in one embodiment the data set is processed prior to step I) using a statistical method to remove outliers. Examples for suitable statistical methods include variance analysis, mutual information and Z-score. Outliers can be removed on the basis of values such as medians or expected values derived from the aforementioned methods. The dimensions of the feature space created by recipe details, atmospheric conditions, process details and properties of the final product may also be reduced to the essential features. The statistical analysis also permits to identify common features or clusters in the data, for example by using scatter plots or principal component methods.

In another embodiment step II) comprises: splitting the data set into a training data set and a testing data set, generating each regression model using the training data set, and validating the regression model against the testing data set.

Candidates for regression models which are validated against the testing data set may be cross-validated according to criteria such as the coefficient of determination R ² , the mean absolute error MAE or the root mean square error RMSE. The regression model for each target physical property with the highest R ² and the lowest MAE and RMSE values can then be selected.

In another embodiment the data set is grouped into clusters and each cluster is split into a training data set and a testing data set for generating the regression models.

In another embodiment at least one of steps I) to VI) comprises providing an interaction with a user via a graphical user interface. For example, the calculated formulas for the formulation parameters can be integrated in a user interface, where the plant technician inserts its standard formulation. In the same time the climate condition is measured by sensors implemented in the production plant and foam storage area. The measured information with the standard formulation of the plant technician are consumed by the formulas, which then calculates the climate corrected formulation. The climate corrected formulation is visualized in the user interface, where the plant technician can manually correct the set points.

In another scenario the calculated formulas are implemented in a FPGA, which controls the set points of the machinery. The plant technician inserts its standard formulation in a user interface. The inserted formulation with the measured climate information of the above mentioned sensors are consumed by the FPGA. The FPGA can now change the set points of the machinery without the interaction with the plant technician.

In another embodiment steps V) to VIII) occur in 60 seconds or less. In another embodiment at least two target physical properties are considered.

In another embodiment at least two climate parameters are considered.

In another embodiment the at least one climate parameter at the production site is the air pressure, the air temperature, the air’s relative humidity or a combination thereof.

In another embodiment the at least one target physical property is the raw density according to DIN EN ISO 845 and/or the compression load deflection at 40% compression according to EN ISO 3386.

In another embodiment the chemical product is a polyurethane foam and the reaction mixture comprises a polyisocyanate, a polyisocyanate-reactive compound and a blowing agent. Examples for suitable polyisocyanates include TDI, MDI, H12-MDI, XDI, HDI, IPDI and their derivatives such as oligomers, NCO-terminated prepolymers, biurets, uretdiones, allophanates and isocyanurates. Examples for suitable polyols include polyester polyols, polyester polyols, polycarbonate polyols, polyetherester polyols and polyestercarbonate polyols. Blowing agents can be physical blowing agents such as cyclopentane or chemical blowing agents such as water or formic acid.

In another embodiment composing the reaction mixture in step VII) comprises, starting from a pre-defined standard reaction mixture composition:

(a) increasing or decreasing a molar ratio of isocyanate groups to isocyanate-reactive groups;

(b) increasing or decreasing an amount of blowing agent;

(c) increasing or decreasing relative amounts of physical and chemical blowing agents with respect to each other; or

(d) a combination thereof.

In another embodiment the method further comprises:

IX) measuring, at the production site, the at least one climate parameter used in the combined regression model after step VIII has begun;

X) calculating the composition of the reaction mixture or process conditions of the reaction using the combined regression model and the at least one measured climate parameter in step

IX; XI) adjusting, or deciding not to adjust, the reaction mixture or process conditions according to the calculation in step X; and

XII) reacting the reaction mixture to obtain the chemical product after making the adjustment or the decision not to adjust, the reaction mixture or process conditions in step XL

It is preferred that steps IX to XII are repeated at least once every 60 seconds, at least once every 30 seconds, or continuously.

In another embodiment the method further comprises:

IX’) measuring, at the production site, the at least one climate parameter used in the combined regression model after step VIII has begun;

X’) determining if the measurement in step IX’ is different from a previous measurement of the at least one climate parameter;

CG) if the measurement has been determined to be different, then calculating the composition of the reaction mixture or process conditions of the reaction using the combined regression model and the at least one measured climate parameter in step IX’ ;

CIG) adjusting the reaction mixture or process conditions according to the calculation in step CG , if such a calculation is made; and

CIIG) reacting the reaction mixture to obtain the chemical product after making the adjustment, the reaction mixture or process conditions in step CIG, if a calculation is made in step XL

It is preferred that steps IX’ to XL are repeated at least once every 60 seconds, at least once every 30 seconds, or continuously.

It is also preferred that at least two climate parameters are measured in steps V and IX’ .

The present invention is furthermore directed towards a system configured for producing a chemical product from a reaction mixture containing at least two components, the system comprising: one or more hardware processors configured by machine-readable instructions to: provide a data set comprising a plurality of production instances, each production instance comprising data on the composition of the reaction mixture, at least one climate parameter at the production site of the product, and at least one target physical property of the product; generate a regression model for each target physical property of the product using the data set, wherein in the regression model a respective target physical property is the dependent variable and other data in the data set are independent variables; transform each regression model to obtain a transformed regression model where the composition of the reaction mixture and/or the process conditions of a reaction depends on at least one climate parameter at the production site of the product and each target physical property is assigned a desired value; combine the transformed regression models to obtain a combined regression model; measuring, at the production site, at least one climate parameter used in the combined regression model; calculate the composition of the reaction mixture and or the process conditions for the reaction using the combined regression model and the at least one measured climate parameter; and compose the reaction mixture and or process conditions according the calculation; and a reactor for reacting the reaction mixture to obtain the chemical product.

Another aspect of the present invention is a computing platform configured for producing a chemical product from a reaction mixture containing at least two components, the computing platform comprising: a non-transient computer-readable storage medium having executable instructions embodied thereon; and one or more hardware processors configured to execute the instructions to: provide a data set comprising a plurality of production instances, each production instance comprising data on the composition of the reaction mixture, at least one climate parameter at the production site of the product, and at least one target physical property of the product; generate a regression model for each target physical property of the product using the data set, wherein in the regression model a respective target physical property is the dependent variable and other data in the data set are independent variables; transform each regression model to obtain a transformed regression model where the composition of the reaction mixture and/or the process conditions of a reaction depends on at least one climate parameter at the production site of the product and each target physical property is assigned a desired value; combine the transformed regression models to obtain a combined regression model; measuring, at the production site, at least one climate parameter used in the combined regression model; calculate the composition of the reaction mixture and or the process conditions for the reaction using the combined regression model and the at least one measured climate parameter; and compose the reaction mixture and or process conditions according the calculation.

Preferably the computing platform is in data communication with a reactor for reacting the reaction mixture to obtain the chemical product.

The present invention will be further described with reference to the following figures without wishing to be limited by them.

FIG. 1 illustrates a system configured for producing a chemical product from a reaction mixture containing at least two components, in accordance with one or more implementations.

FIG. 2 illustrates a method for producing a chemical product from a reaction mixture containing at least two components, in accordance with one or more implementations.

FIG. 1 illustrates a system 100 configured for producing a chemical product from a reaction mixture containing at least two components, in accordance with one or more implementations. In some implementations, system 100 includes one or more servers 102. Server(s) 102 is configured to communicate with one or more client computing platforms 104 according to a client/server architecture and/or other architectures. Client computing platform(s) 104 is configured to communicate with other client computing platforms via server(s) 102 and/or according to a peer-to-peer architecture and or other architectures. Users may access system 100 via client computing platform(s) 104.

Server(s) 102 is configured by machine-readable instmctions 106. Machine-readable instructions 106 include one or more instruction modules. The instruction modules include computer program modules. The instruction modules include one or more of a data providing module 108, a regression model generating module 110, a regression model transformation module 112, a regression model combining module 114, a composition calculation module 118, a reaction mixture composition module 120, a reaction mixture reaction module 122, a climate parameter measuring module 124, a reaction mixture process condition adjusting module 126, a reaction mixture process condition adjusting module 128, a measurement determination module 130, and or other instruction modules.

Data providing module 108 is configured to provide a data set including a plurality of production instances. By way of non-limiting example, each production instance includes data on the composition of the reaction mixture, at least one climate parameter at the production site of the product, and at least one target physical property of the product. The data set is processed prior to providing the data set using a statistical method to remove outliers. The at least one target physical property is the raw density according to DIN EN ISO 845 and or the compression load deflection at 40 % compression according to EN ISO 3386.

Regression model generating module 110 is configured to generate a regression model for each target physical property of the product using the data set. Generating the regression model includes splitting the data set into a training data set and a testing data set. Generating the regression model includes generating each regression model using the training data set. Generating the regression model includes validating the regression model against the testing data set. The data set is grouped into clusters and each cluster is split into a training data set and a testing data set for generating the regression models.

In the regression model a respective target physical property is the dependent variable and other data in the data set are independent variables.

Regression model transformation module 112 is configured to transform each regression model to obtain a transformed regression model where the composition of the reaction mixture and or the process conditions of a reaction depends on at least one climate parameter at the production site of the product and each target physical property is assigned a desired value. By way of non-limiting example, the at least one climate parameter at the production site is the air pressure and/or the air temperature.

Regression model combining module 114 is configured to combine the transformed regression models to obtain a combined regression model.

Composition calculation module 118 is configured to calculate the composition of the reaction mixture and/or the process conditions for the reaction using the combined regression model and the at least one measured climate parameter. .

Composition calculation module 118 is alternatively configured to, if the measurement has been determined to be different, calculate the composition of the reaction mixture or process conditions of the reaction using the combined regression model and the at least one measured climate parameter in measuring the at least one climate.

Reaction mixture composition module 120 is configured to compose the reaction mixture and/or process conditions according the calculation.

Reaction mixture reaction module 122 is configured to react the reaction mixture to obtain the chemical product after making the adjustment. The reaction mixture or process conditions in may adjust the reaction mixture if a calculation is made in adjustinging the reaction.

Climate parameter measuring module 124 is configured to measuring, at the production site, the at least one climate parameter used in the combined regression model before, during and or after reacting the reaction mixture has begun.

Reaction mixture process condition adjusting module 126 is configured to adjust the reaction mixture or process conditions according to the calculation in calculating the composition.

Measurement determination module 130 is configured to determine if the measurement in measuring the at least one climate is different from a previous measurement of the at least one climate parameter.

In some implementations, at least two target physical properties are considered. In some implementations, at least two climate parameters are considered. In some implementations, the air's temperature, pressure or relative humidity or a combination thereof are considered. In some implementations, by way of non-limiting example, the chemical product is a polyurethane foam and the reaction mixture includes a polyisocyanate, a polyisocyanate- reactive compound and a blowing agent. In some implementations, composing the reaction mixture includes starting from a pre-defined standard reaction mixture composition increasing or decreasing a molar ratio of isocyanate groups to isocyanate-reactive groups.

In some implementations, composing the reaction mixture includes starting from a pre defined standard reaction mixture composition increasing or decreasing an amount of blowing agent. In some implementations, composing the reaction mixture includes starting from a pre-defined standard reaction mixture composition increasing or decreasing relative amounts of physical and chemical blowing agents with respect to each other.

In some implementations, server(s) 102, client computing platform(s) 104, and/or external resources 132 are operatively linked via one or more electronic communication links. For example, such electronic communication links is established, at least in part, via a network such as the Internet and or other networks. It will be appreciated that this is not intended to be limiting, and that the scope of this disclosure includes implementations in which server(s) 102, client computing platform(s) 104, and or external resources 132 is operatively linked via some other communication media.

A given client computing platform 104 includes one or more processors configured to execute computer program modules. The computer program modules is configured to enable an expert or user associated with the given client computing platform 104 to interface with system 100 and or external resources 132, and/or provide other functionality attributed herein to client computing platform(s) 104. By way of non-limiting example, the given client computing platform 104 includes one or more of a desktop computer, a laptop computer, a handheld computer, a tablet computing platform, a NetBook, a Smartphone, a gaming console, and or other computing platforms.

External resources 132 include sources of information outside of system 100, external entities participating with system 100, and or other resources. In some implementations, some or all of the functionality attributed herein to external resources 132 is provided by resources included in system 100.

Server(s) 102 include electronic storage 134, one or more processors 136, and/or other components. Server(s) 102 includes communication lines, or ports to enable the exchange of information with a network and/or other computing platforms. Illustration of server(s) 102 in FIG. 1 is not intended to be limiting. Server(s) 102 include a plurality of hardware, software, and or firmware components operating together to provide the functionality attributed herein to server(s) 102. For example, server(s) 102 are implemented by a cloud of computing platforms operating together as server(s) 102.

Electronic storage 134 may comprise non-transitory storage media that electronically stores information. The electronic storage media of electronic storage 134 includes one or both of system storage that is provided integrally (i.e., substantially non-removable) with server(s) 102 and/or removable storage that is removably connectable to server(s) 102 via, for example, a port (e.g., a USB port, a firewire port, etc.) or a drive (e.g., a disk drive, etc.). Electronic storage 134 includes one or more of optically readable storage media (e.g., optical disks, etc.), magnetically readable storage media (e.g., magnetic tape, magnetic hard drive, floppy drive, etc.), electrical charge-based storage media (e.g., EEPROM, RAM, etc.), solid- state storage media (e.g., flash drive, etc.), and/or other electronically readable storage media. Electronic storage 134 includes one or more virtual storage resources (e.g., cloud storage, a virtual private network, and or other virtual storage resources). Electronic storage 134 may store software algorithms, information determined by processor(s) 136, information received from server(s) 102, information received from client computing platform(s) 104, and or other information that enables server(s) 102 to function as described herein.

Processor(s) 136 is configured to provide information processing capabilities in server(s) 102. As such, processor(s) 136 includes one or more of a digital processor, an analog processor, a digital circuit designed to process information, an analog circuit designed to process information, a state machine, and/or other mechanisms for electronically processing information. Although processor(s) 136 is shown in FIG. 1 as a single entity, this is for illustrative purposes only. In some implementations, processor(s) 136 includes a plurality of processing units. These processing units is physically located within the same device, or processor(s) 136 may represent processing functionality of a plurality of devices operating in coordination. Processor(s) 136 is configured to execute modules 108, 110, 112, 114, 118, 120, 124, 126 and or 130, and or other modules by software, hardware, firmware, some combination of software, hardware, and/or firmware; and/or other mechanisms for configuring processing capabilities on processor(s) 136. As used herein, the term “module” may refer to any component or set of components that perform the functionality attributed to the module. This includes one or more physical processors during execution of processor readable instructions, the processor readable instructions, circuitry, hardware, storage media, or any other components.

It should be appreciated that although modules 108, 110, 112, 114, 118, 120, 124, 126 and/or 130 are illustrated in FIG. 1 as being implemented within a single processing unit, in implementations in which processor(s) 136 includes multiple processing units, one or more of modules 108, 110, 112, 114, 118, 120, 124, 126 and/or 130 is implemented remotely from the other modules. The description of the functionality provided by the different modules 108, 110, 112, 114, 118, 120, 124, 126 and or 130 described below is for illustrative purposes, and is not intended to be limiting, as any of modules 108, 110, 112, 114, 118, 120, 124, 126 and or 130 may provide more or less functionality than is described. For example, one or more of modules 108, 110, 112, 114, 118, 120, 124, 126 and/or 130 is eliminated, and some or all of its functionality is provided by other ones of modules 108, 110, 112, 114, 118, 120, 124, 126 and/or 130. As another example, processor(s) 136 is configured to execute one or more additional modules that may perform some or all of the functionality attributed below to one of modules 108, 110, 112, 114, 118, 120, 124, 126 and or 130.

FIG. 2 illustrates a method 200 for producing a chemical product from a reaction mixture containing at least two components, in accordance with one or more implementations. The operations of method 200 presented below are intended to be illustrative. In some implementations, method 200 is accomplished with one or more additional operations not described, and/or without one or more of the operations discussed.

In some implementations, method 200 is implemented in one or more processing devices (e.g., a digital processor, an analog processor, a digital circuit designed to process information, an analog circuit designed to process information, a state machine, and/or other mechanisms for electronically processing information). The one or more processing devices includes one or more devices executing some or all of the operations of method 200 in response to instructions stored electronically on an electronic storage medium. The one or more processing devices includes one or more devices configured through hardware, firmware, and or software to be specifically designed for execution of one or more of the operations of method 200.

An operation 202 includes providing a data set including a plurality of production instances. Each production instance includes data on the composition of the reaction mixture, at least one climate parameter at the production site of the product, and at least one target physical property of the product. Operation 202 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to data providing module 108, in accordance with one or more implementations.

An operation 204 includes generating a regression model for each target physical property of the product using the data set. In the regression model a respective target physical property is the dependent variable and other data in the data set are independent variables. Operation 204 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to regression model generating module 110, in accordance with one or more implementations.

An operation 206 includes transforming each regression model to obtain a transformed regression model where the composition of the reaction mixture and/or the process conditions of a reaction depends on at least one climate parameter at the production site of the product and each target physical property is assigned a desired value. Operation 206 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to regression model transformation module 112, in accordance with one or more implementations.

An operation 208 includes combining the transformed regression models to obtain a combined regression model. Operation 208 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to regression model combining module 114, in accordance with one or more implementations.

An operation 210 includes measuring, at the production site, at least one climate parameter used in the combined regression model. Operation 210 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to Measuring module 116, in accordance with one or more implementations.

An operation 212 includes calculating the composition of the reaction mixture and or the process conditions for the reaction using the combined regression model and the at least one measured climate parameter. Operation 212 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to composition calculation module 118, in accordance with one or more implementations.

An operation 214 includes composing the reaction mixture and/or process conditions according the calculation. Operation 214 is performed by one or more hardware processors configured by machine-readable instructions including a module that is the same as or similar to reaction mixture composition module 120, in accordance with one or more implementations. An operation 216 includes reacting the reaction mixture to obtain the chemical product. Operation 216 is performed by one or more hardware processors configured by machine- readable instructions including a module that is the same as or similar to reaction mixture reaction module 122, in accordance with one or more implementations. The present invention will be further described in the following section without wishing to be limited by the proceedings outlined therein. A dataset having the structure given below may be used. Each row represents data for one production instance of polyurethane foam. The foam may be produced by reacting a polyol component with a polyisocyanate component. The data may be normalized as standard scores (z = (c-m)/s)). Water: amount of water [g] added to the polyol component as a chemical blowing agent;

ISO: NCO index of the reaction mixture; AbsHum_Mean: mean value for the absolute air humidity [g/m ³ ] at the production site; p: air pressure [kPa] at the production site; RD: raw density of the polyurethane foam [kg/m ³ ]; CH40: compression load deflection of the polyurethane foam at 40% compression [kPa].

In alignment with the method according to the invention each row of the table represents a production instance. The composition of the reaction mixture, as it pertains to the method according to the invention, is documented in the columns “ISO” and “Water”. Likewise, climate parameters are “AbsHum_Mean” and “p” and target physical properties are “RD” and “CH40”.

For the raw density “RD” the following regression model may be generated:

RD = constant_01

+ constant_02 · XI + constant_03 · X2 + constant_04 · X4

In accordance with the method according to the invention the raw density is the dependent variable and a part of the other data - ISO, water and air pressure - are the independent variables. The absolute humidity has not been taken into account. Furthermore, also in accordance with the method according to the invention, the compression load deflection as the other target physical property has been excluded.

For the compression hardness “CH40” the following regression model was generated:

CH40 = constant_05

+ constant_06 · XI + constant_07 · X2 + constant_08 · X3 + constant_09 · X4 + constant_10 - XI ² + constant_l 1 · X2 ² + constant_12 - XI · X2

In accordance with the method according to the invention the compression load deflection is the dependent variable and the other data - ISO, water, absolute humidity and air pressure - are the independent variables. Furthermore, also in accordance with the method according to the invention, the raw density as the other target physical property has been excluded.

The regression models were transformed as follows. The subscript “_nr” denotes “new recipe”, i.e relating to the recipe for the polyurethane foam after taking climate corrections (absolute humidity and air pressure) into account. The constants constant_01 to constant_33 may be positive or negative and have small values, usually from -2 to 2.

RD - RD_nr = 0; in expanded form:

(constant_01 + constant_02 · XI + constant_03 · X2 + constant_04 · X4)

- (constant_01 + constant_02 · Xl_nr + constant_03 · X2_nr + constant_04 · X4_nr)

= 0 where RD_nr has the new climate-corrected components Xl_nr, X2_nr and X4_nr.

CH40 - CH40_nr = 0; in expanded form: (constant_05 + constant_06 · XI + constant_07 · X2 + constant_08 · X3 + constant_09 · X4 + constant_10 - XI ²

+ constant_l 1 · X2 ² + constant_12 - XI · X2)

- (constant_05 + constant_06 · Xl_nr + constant_07 · X2_nr

+ constant_08 · X3_nr + constant_09 · X4_nr + constant_10 · Xl_nr ² + constant_l 1 · X2_nr ² + constant_12 · Xl_nr · X2_nr)

= 0 where CH40_nr has also the new climate-corrected components Xl_nr, X2_nr and X4_nr. The transformed regression models were combined as follows (sqrt: quare root):

XI nr = constant_13

+ constant_14 · X2 + constant_14 · X4_nr + constant_15 · XI - constant_14 · X4 + constant_16 · sqrt( constant_17 + constant_18 · X4_nr + constant_19 · X2 + constant_20 · X3_nr

- constant_18 · X4

- constant_21 · XI

- constant_20 · X3

+ constant_22 · X2 · X4_nr + constant_23 · XI · X4_nr

- constant_22 · X2 · X4

- constant_23 - XI · X4 - constant_24 · X4 · X4_nr

- constant_25 · XI · X2 + constant_26 - XI ²

+ constant_27 · X4 ² + constant_27 · X4_nr ²

+ constant_28 · X2 ² )

X2 nr = - constant_29

+ constant_30 · X2 + constant_31 - XI + constant_32 · X4 _nr

- constant_32 · X4

- constant_33 · sqrt( constant_17 + constant _^ 18 · X4_nr + constant_19 · X2 + constant_20 · X3_nr

- constant_18 · X4

- constant_21 · XI

- constant_20 · X3

+ constant_22 · X2 · X4_nr + constant_23 · XI · X4_nr

- constant_22 · X2 · X4

- constant_23 - XI · X4

- constant_24 · X4 · X4_nr

- constant_25 · XI · X2 + constant_26 - XI ²

+ constant_27 · X4 ² + constant_27 · X4_nr ² + constant_28 · X2 ² )

X3 nr and X4_nr are the new measured absolute humidity and air pressure values. X3 and X4 are the standard lab climate conditions.