Tetra Laval Holdings & Finance

There is described a package forming apparatus (1) comprising a first operative unit (4) and a second operative unit (5) each of which comprises a plurality of operative devices (7) and a conveyor device (8) configured to advance the plurality of operative devices (7). Each operative device (7) comprises a cart (10) moveably coupled to the conveyor device (8) and an operative group (11). Each conveyor device (8) comprises a control unit (12) configured to advance the carts (10) along an advancement path (Q) and a braking device (18) configured to interact with a respective braking pad (19) of each operative device (7) and configured to allow and to impede advancement of the operative devices (7). Each braking device (18) comprises a first plate assembly (20) and a second plate assembly (21) spaced apart from one another, forming a space (22) therebetween and being movable between a release position and a clamping position. The actuator assembly (23) comprises at least one actuator (30) connected to the first plate assembly (20) and configured to induce a movement of the first plate assembly (20). The at least one actuator (30) comprises a piston (31) and a ball joint assembly (35) connected to the piston (31) and the first plate assembly (20).

A method (200) for producing a diary product (P1). The method (200) comprising: feeding (S202) a dairy material (DM) to a first filtration stage (101); filtering (S204) the dairy material (DM), in the first filtration stage (101), into a first permeate stream (P1) and a first retentate stream (R1); feeding (S206) the first retentate stream (R1) to a second filtration stage (102); filtering (S208) the first retentate stream (R1) into a second permeate stream (P2) and a second retentate stream (R2); recirculating (S210) the second permeate stream (P2) by feeding it to the first filtration stage (101); and collecting (S212) the first permeate stream (P1) as the dairy product. The disclosure further relates to a system (100) configured to produce a dairy product.

The present disclosure relates to a method (400) for producing a tea-based beverage precursor (208). The method comprises obtaining (402) brew spent tea leaves (BSTL) (201) from a tea extraction system (102) used in the production of tea. These BSTL (201) are then subjected to a wet grinding process (404) to form a BSTL slurry (704). Subsequently, the BSTL slurry (704) is processed (406) using enzymatic treatment to convert it into the tea-based beverage precursor (208). Related products, processes and processing lines are also disclosed.

There is described a method for producing sealed packages (2) containing a pourable product, the method comprises the steps of: a) advancing a web (4) of packaging material along an advancing path (P); b) folding the web (4) into a tube (3); c) filling the tube (3) with the pourable product by means of a filling device (7); d) detecting a deviation or anomaly at the web (4) indicative of the state of the packaging material; e) adjusting the operation of the filling device (7) as a function of the detection.

The present invention relates to the use of a paper substrate (11) in a high-quality, barrier-coated paper substrate (10) for providing good toughness, flexibility and oxygen barrier properties to a laminated packaging material and to packaging containers comprising the laminated packaging material. The barrier-coated paper substrate may also provide heat sealing capability to a laminated packaging material comprising it. The barrier-coated paper substrate (10; 23), is suitable in packaging of oxygen-sensitive food products.

A computer-implemented method (900) for identifying packages (116) to be discarded and sample packages (302) for quality assessment in a package flow (300) is disclosed. The method comprises obtaining (902) a package ID related to one of the packages of the package flow (300), obtaining (904) sensor data (810) associated to the package ID, applying (906) a multi-output model designed for multitask learning on the sensor data (810), wherein the multi-output model is configured to output a discarding prediction as well as a sampling prediction, forming (908) a decision whether to remove the package from the package flow (300), pull the package from the package flow (300) to conduct quality assessment, or allow the package to proceed in the package flow (300) based on the discarding prediction and the sampling prediction, in case the decision (910) is to remove the package or pull the package from the package flow (300), generating (912) control data (816, 818) reflecting the decision, wherein the control data (816, 818) comprises command signals, and transmitting (914) the control data (816, 818) to a reject mechanism (820) arranged for removing the package from the conveyor system (308) or to a sampling mechanism (824) arranged for pulling the package from the package flow (300) to a quality assessment station (814).

The present invention relates to a cleaning system (10) for cleaning a filling machine configured to fill packages with a liquid food product. The cleaning system comprises a cleaning module (100), a foam supply module (200), a water supply module (300) and a water pump (400). The water pump is configured to pressurize the water supplied to the cleaning module or to the foam supply module from the water supply module. The cleaning system further comprises a control unit (500) configured to control an operational state of the water pump based on a reading from a water flow meter (310) in the water supply module or on a reading from an air flow meter (230) of the foam supply module. Figure elected for publication: Fig. 1

A method for monitoring a transversal sealing process 200 involves receiving measurement data 304 from a packaging line 100. A trained machine-learning model 302, which is fed with the measurement data 304, generates a quality indicator 306 that reflects the status of at least one package 102's transversal sealing 104. This approach allows real-time quality control and enhances productivity by identifying potential issues or deviations from the standard sealing process.

The present invention relates to training a machine learning model to predict a relative crack index, RCI, map for a candidate fiberboard patterning design to be used in patterning of fiberboard packaging material. The training comprises: generating a training dataset for the machine learning model by, for each of a plurality of training fiberboard patterning designs, performing a finite element method simulation resulting in a respective training RCI map; and training the machine learning model by adjusting weights of the machine learning model using the training fiberboard patterning designs of the training dataset as input and the training RCI maps of the training dataset as ground truth. A training apparatus for training the machine learning model is also presented as well as use of the machine learning model to predict a RCI map for a candidate fiberboard patterning design.

The present invention relates to training a machine learning model to predict a relative crack index, RCI, map for a candidate combination of a paperboard patterning design and a paperboard material having a set of paperboard material model parameters, the candidate combination to be used in patterning of paperboard packaging material. The training comprises: generating a training dataset for the machine learning model by, for each training combination of one or more training paperboard patterning designs and one or more paperboard materials having a respective set of paperboard material model parameters, performing a finite element method simulation resulting in a respective training RCI map; and training the machine learning model by adjusting weights of the machine learning model using the training combinations of the training dataset as input and the training RCI maps of the training dataset as ground truth. A training apparatus for training the machine learning model is also presented as well as use of the machine learning model to predict a RCI map for a candidate paperboard patterning design.