This thesis was was carried out in collaboration with Logic S.r.l., an automation company based in the province of Pavia, in order to develop an effective and ef- ficient simulation environment and control system for a new automated perfume dosing machine under design. This machine would have at its disposal 600 basic ingredients and by providing the recipe containing the ingredients and their quantities, the machine would take care of dosing them by itself as accurately as possible. The thesis work consisted in studying the mechanical and hydraulic characteristics of the system, in defining a dynamic model of the entire process and implementing it using the Simulink toolbox in order to best replicate the real system. Two control strategies were then implemented: a discrete one with pre-set param- eters, defined by the company, and a proportional control, realized in this thesis. During the thesis, a prototype of the real system was made available for testing purposes and for identification of critical issues related to the new system. The prototype includes a water tank placed on top connected by a tube to a nozzle of adjustable width to dose water into a barrel below. The barrel with the mixture then is weighed on a load cell to obtain information on the dosed amount. The different viscosities of liquids play a crucial role in this type of application since the liquids to be handled have viscosities ranging from that of water to that of honey. In addition, given the variability caused by changes in viscosity due to varying ambient temperature and liquid drop time, the choice for the initial discrete control parameters requires several dosing trials. Therefore, this method- ology is not applicable for all 600 ingredients to be dosed. Creating a simulation environment as faithful as possible to the performance of the real system would, in the first instance, allow the controller parameters to be chosen with a good degree of reliability without resorting in too many dosing tests. In order to collect data on the real system to proceed with the validation of the proposed model, an on-site experience was carried out. The goal was to upgrade the hardware part by connecting an industrial PLC to read the weight with the load cell and to control the dosing valve according to the required algorithm. This was followed by collecting both discrete and proportional control data on the real system. When the results were compared with those of the simulation, a considerable and unsatisfactory discrepancy was found between the simulation and experimental data. Subsequent analysis revealed that this was due to the lack of modeling of measurement error. Therefore, data analysis, literature study and disturbance modeling were carried out to create a more realistic simulation. Comparison between the experimental data and the new version of the simula- tor showed a good match that allowed the proportional control to be calibrated appropriately. With this calibration developed in simulation, it was possible to obtain dosing errors well below the company’s expectations.
Progettazione e implementazione di un sistema di controllo per il dosaggio di prodotti chimici e cosmetici. Questa tesi `e stata `e stata svolta in collaborazione con Logic S.r.l., un’azienda che si occupa di automazione ed ha sede in provincia di Pavia, al fine di sviluppare una ambiente di simulazione e un sistema di controllo efficace ed efficiente per una nuova macchina per il dosaggio automatizzato di profumi in fase di progettazione. Questa macchina avrebbe a disposizione 600 ingredienti di base e fornendo la ricetta contenente gli ingredienti e le loro quantit`a, la macchina si dovrebbe oc- cupare da sola di dosarli nel modo pi `u preciso possibile. Il lavoro di tesi `e consistito nello studio delle caratteristiche meccaniche ed idrauliche del sistema, nella definizione di un modello dinamico dell’intero processo e della sua implementazione tramite il toolbox Simulink al fine di replicare al meglio il sistema reale. In seguito sono state implementate due strategie di controllo: una discreta con parametri pre-impostati, definito dall’azienda, e un controllo proporzionale, real- izzato in questa tesi. Durante la tesi, `e stato reso disponibile un prototipo del sistema vero a fini di test e per l’identificazione delle criticit`a legate al nuovo sistema. Il prototipo comprende un serbatoio di acqua posto in alto collegato con un tubo ad un ugello di ampiezza regolabile per dosare l’acqua in un barile sottostante. Il barile con la miscela poi `e pesato su una cella di carico per ottenere l’informazione sulla quantit`a dosata. La diversa viscosit`a dei liquidi gioca un ruolo determinante in questo tipo di appli- cazione dato che i liquidi da gestire hanno viscosit`a comprese tra quella dell’acqua e quella del miele. Considerata inoltre la variabilit`a causata da cambi di viscosit`a dovuti ad una temperatura ambiente variabile e dal tempo di caduta del liquido, la scelta dei parametri del controllo discreto iniziale richiede diverse prove di dosag- gio. Questa metodologia non risulta quindi applicabile per tutti e 600 gli ingre- dienti da dosare. La creazione di un ambiente di simulazione il pi `u possibile fedele all’andamento del sistema reale, permetterebbe in prima istanza di scegliere i parametri del controllore con un buon grado di affidabilit`a senza ricorrere a troppi test di dosaggio. Al fine di raccogliere dati sul sistema reale per procedere alla validazione del mod- ello proposto, `e stata svolta un’esperienza in azienda. L’obiettivo era di potenziare la parte hardware collegando un PLC industriale per leggere il peso con la cella di carico e per comandare la valvola di dosaggio secondo l’algoritmo richiesto. Si `e quindi proceduto con raccogliere i dati del controllo sia discreto che proporzionale sul sistema vero. Messi a confronto i risultati con quelli della simulazione, `e stata rilevata una dis- crepanza notevole e non soddisfacente fra la simulazione e i dati sperimentali. Successive analisi hanno rivelato che ci `o fosse dovuto alla mancanza di model- lizzazione di un errore di misura. Si `e proceduto quindi con l’analisi dei dati, lo studio della letteratura e alla modellizzazione del disturbo, in modo da creare un simulazione pi `u realistico. Il confronto fra i dati sperimentali e la nuova versione del simulatore ha evidenzi- ato una buona corrispondenza che ha permesso di tarare il controllo proporzionale in maniera opportuna. Con questa taratura sviluppata in simulazione `e stato pos- sibile ottenere errori di dosaggio ben inferiori alle aspettative dell’azienda
Design and implementation of a control system for dosing of chemicals and cosmetics products.
MONGINI, PAOLO ALBERTO
2022/2023
Abstract
This thesis was was carried out in collaboration with Logic S.r.l., an automation company based in the province of Pavia, in order to develop an effective and ef- ficient simulation environment and control system for a new automated perfume dosing machine under design. This machine would have at its disposal 600 basic ingredients and by providing the recipe containing the ingredients and their quantities, the machine would take care of dosing them by itself as accurately as possible. The thesis work consisted in studying the mechanical and hydraulic characteristics of the system, in defining a dynamic model of the entire process and implementing it using the Simulink toolbox in order to best replicate the real system. Two control strategies were then implemented: a discrete one with pre-set param- eters, defined by the company, and a proportional control, realized in this thesis. During the thesis, a prototype of the real system was made available for testing purposes and for identification of critical issues related to the new system. The prototype includes a water tank placed on top connected by a tube to a nozzle of adjustable width to dose water into a barrel below. The barrel with the mixture then is weighed on a load cell to obtain information on the dosed amount. The different viscosities of liquids play a crucial role in this type of application since the liquids to be handled have viscosities ranging from that of water to that of honey. In addition, given the variability caused by changes in viscosity due to varying ambient temperature and liquid drop time, the choice for the initial discrete control parameters requires several dosing trials. Therefore, this method- ology is not applicable for all 600 ingredients to be dosed. Creating a simulation environment as faithful as possible to the performance of the real system would, in the first instance, allow the controller parameters to be chosen with a good degree of reliability without resorting in too many dosing tests. In order to collect data on the real system to proceed with the validation of the proposed model, an on-site experience was carried out. The goal was to upgrade the hardware part by connecting an industrial PLC to read the weight with the load cell and to control the dosing valve according to the required algorithm. This was followed by collecting both discrete and proportional control data on the real system. When the results were compared with those of the simulation, a considerable and unsatisfactory discrepancy was found between the simulation and experimental data. Subsequent analysis revealed that this was due to the lack of modeling of measurement error. Therefore, data analysis, literature study and disturbance modeling were carried out to create a more realistic simulation. Comparison between the experimental data and the new version of the simula- tor showed a good match that allowed the proportional control to be calibrated appropriately. With this calibration developed in simulation, it was possible to obtain dosing errors well below the company’s expectations.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/16707