Abstract This thesis pursues to evidence that it is possible to get information on the behaviour of a complex chemical process by data-based analysis, even though the accurate reaction chemistry is not known. In the hydrometallurgical zinc production process, metallic zinc is produced by reducing it from aqueous solution. The essential part of the process is the purification of zinc sulphate solution, where all elements nobler than zinc are removed by deposition from the solution. These elements lower the efficiency of the electric current drastically and cause zinc dissolution even in the smallest amounts. Process monitoring supplies a remarkable amount of on-line measurement data and analysis information. This makes it possible to use data-based methods for the evaluation of deposition reactions. The purpose of this thesis was not to specify reaction equations, but to evaluate which of the known reactions will occur in different process conditions. This was done by diagnosing the purification process of zinc sulphate solution and by combining measurement information with reaction models. An increased understanding of process stability also affects solution purification costs. The results gave new information about the purification process of zinc sulphate solution and its accelerating and decelerating components. The second target was to develop a modular, model-free method of signal validation and estimation, suitable for implementation in the control system in the form of a simple, configurable algorithm. Combining measurements confidence level-based information with fuzzy logic provides a compact system that is easily implemented in the process automation system. The algorithm that was developed was tested in a direct leaching process, where the combination of fast on-line sensors and accurate reference measurement was needed. Measurement data came from an operational zinc plant, so the amount of factory testing was strictly limited. Data evaluation was based on long-term variations in process conditions, possible disturbances and breakdowns of the measurement devices. / Tiivistelmä Tämä väitöskirja pyrkii osoittamaan, että monimutkaisen kemiallisen prosessin toiminnasta voidaan saada tietoa prosessimittauksista laskettujen tunnuslukujen ja niihin vaikuttavien tekijöiden mallintamisen avulla, vaikka reaktioiden tarkkaa kemiaa ei tunneta. Tutkimuskohteena käytetyssä sinkin hydrometallurgisessa valmistuksessa tärkeimpiä osaprosesseja ovat liuotus, haitta-aineiden saostus ja elektrolyyttinen pelkistys. Haitta-aineiden saostaminen tapahtuu reagenssien avulla. Reaktiossa on mukana sekä kiinteä, nestemäinen ja kaasufaasi, joiden välisiä reaktiota ja yhteisvaikutusta ei pystytä tarkasti tutkimaan ja määrittämään. Prosessista on kuitenkin saatavilla huomattava määrä mittaus- ja analyysitietoa, joten eri saostusreaktioihin vaikuttavien tekijöiden vaihtelua voidaan tutkia datapohjaisesti. Tutkimuksen tarkoituksena oli selvittää, kuinka suurelta osin tunnetut reaktiot toteutuvat ja kuinka suuri on se osuus reaktioista, joita ei pystytä selittämään. Lisäksi eri luotettavuusasteisen mittaustiedon yhdistämisellä pyrittiin lisäämään mittaustiedon luotettavuutta ja tuottamaan operaattoreille entistä käyttökelpoisempaa informaatiota prosessin tilasta. Tätä varten kehitettiin helposti käyttökohteen mukaan adaptoitava laskenta-algoritmi, joka tuottaa mittaustietoihin perustuvan estimaatin. Estimaatin toimintaa testattiin suoraliuotusprosessissa, josta oli saatavilla sekä on-line mittauksia että tarkkoja referenssimittauksia. Tutkimusaineistona käytettiin Kokkolan sinkkitehtaalta saatavaa mittausdataa. Mittausaineisto työhön on suurimmaksi osaksi peräisin toimivasta tehdasprosessista, joten tuotannolliset seikat ovat voimakkaasti rajoittaneet prosessikokeiden tekoa. Työssä onkin turvauduttu pitkällä aikavälillä tapahtuneiden muutosten sekä mahdollisten prosessihäiriöiden ja analysaattorien vikaantumisten aiheuttamien muutosten analysointiin. Tilastollisen analysoinnin avulla selvitettiin, tapahtuuko prosessiparametrien välisissä riippuvuuksissa muutoksia eri mittausaineistojen välillä. Tutkimustulosten perusteella pystytään pitämään saostusprosessi paremmin optimaalisella toiminta-alueella. Kehitetty "optimaalisen estimaatin" laskenta-algoritmi on asennettu toimimaan tehtaan prosessinohjaustietokantaan.

In recent years, a compound N-nitrosodimethylamine (NDMA), a probable human carcinogen, has been identified as an emerging disinfection by-product (DBP) since its formation and detection were linked to chlorine-based disinfection processes in several water utilities in the U.S. and Canada. Numerous organic nitrogen compounds present in water may impact the formation of NDMA during disinfection. Amine-based water treatment polymers used as coagulants and flocculants have been suggested as potential NDMA precursors due to the presence of amine functional groups in their structures, as well as the possible presence of dimethylamine (DMA) residues in polymer products. To minimize the potential risk of NDMA associated with water treatment polymers, the mechanisms of how the polymers behave as NDMA precursors and their contribution to the overall NDMA formation under actual water treatment conditions need to be elucidated. This research involved a systematic investigation to determine whether amine-based water treatment polymers contribute to NDMA formation under drinking water and wastewater treatment conditions, to probe the involved reaction mechanisms, and to develop strategies to minimize the polymers NDMA formation potential. The investigation included five research tasks: (1) General screening of NDMA formation potential of commonly used amine-based water treatment polymers, (2) NDMA formation from amine-based water treatment polymers under relevant water treatment conditions, (3) Probing the mechanisms of NDMA formation from polyamine and PolyDADMAC, (4) Effect of water treatment processes on NDMA formation from amine-based water treatment polymers, and (5) Developing strategies to reduce polymers NDMA formation potential. Direct chloramination or chlorination of high doses of polymers in deionized water at longer than typical contact time was used in the general screening of the NDMA formation potential of water treatment polymers and in the studies to identify reaction mechanisms. On the other hand, realistic dosages of chloramines and polymers and contact time were used in simulating representative water treatment conditions to evaluate the contribution of polymers to the overall NDMA formation in real systems. On the basis of the study results, strategies were developed to reduce the NDMA formation potential of amine-based water treatment polymers, which include modification of polymer structures and treatment parameters.