Forays in Process Analytical Technologies and Theory of Sampling:: 1. Feasibility of handheld NIR characterization in small-scale berry wineries (PAT) 2. Variographic analysis of standard procedures for industrial waste water characterization (TOS)

Bidrag til anvendt Process Analytical Technologies og Theory of Sampling:: 1. Indledende undersøgelse af NIR karakterisering hos en lokal bærvinsproducent 2. Variografisk analyse af standard procedurer for spildevandskarakterisering

Kristensen, Dorthe

4. term

Chemical Engineering, Master

2010

2010-06-14

212

Afhandlingen undersøger to praktiske målemetoder, der kan styrke kvalitetskontrol i små produktioner og forsyningsanlæg. Del 1 viser, at håndholdt nær-infrarød spektroskopi (NIR) – en lysbaseret metode til at estimere indholdet af kemiske stoffer – kan bruges til hurtig og nøjagtig måling af nøglekomponenter i tre lokale, lav-alkoholiske bær- og frugtvine produceret i Tromsø. To Phazir-sensorsystemer (kort- og langbølget NIR) blev anvendt sammen med kemometrisk multivariat kalibrering. For hver vin blev der opbygget og optimeret PLS-regressionsmodeller (partial least squares) for udvalgte komponenter af vinfaglig interesse og valideret med uafhængige testsæt. I den kortbølgede NIR-region (890–1690 nm) gav modellerne tilfredsstillende resultater for saccharose, glukose, fruktose, glycerol, ethanol samt citronsyre, vinsyre, æblesyre, ravsyre, mælkesyre og eddikesyre, og fremragende resultater for total sukker og samlet syreindhold. Førsteafledt forbehandling blev brugt for alle parametre bortset fra ethanol, hvor multiplikativ spredningskorrektion (MSC) blev anvendt. Resultaterne vurderes sandsynligvis at kunne overføres til bordvine og fermenteringsmedier. Del 2 anvender variografisk analyse (Prøvetagningsteori) til at evaluere procesvariation og usikkerheder ved prøvetagning af spildevand ved Bramming Waste Water Treatment Facility North. Der blev analyseret både korttids-inkrementprøver over 48 timer fra to udvalgte punkter og 24-timers kompositprøver for total fosfor (TP), ledningsevne og ammonium (NH4-N). Variogrammet for inkrementerne viste en cyklisk variation på cirka 420 m3, svarende til forskelle i belastning mellem dag og nat. 24-timers kompositprøver blev vurderet som pålidelige til emissionsovervågning. Over 39 dage blev standard kompositprøver (24-timers intervaller) indsamlet fra syv alternative udløbslokationer og sammenlignet. Variografien viste en tydelig 7-dages cyklus for TP og en 9-dages cyklus for ledningsevne, der afspejler uge/weekend-variationer; 9-dages cyklussen forbliver uforklaret. Det standarde prøvetagningspunkt F, umiddelbart efter weiren (overløb), er repræsentativt, og punkt B blev fundet som et ligeværdigt alternativ. Den samlede prøvetagningsfejl (TSE) blev estimeret til 12,3 % for TP og 12,4 % for ledningsevne, mens den samlede analytiske fejl (TAE) var ubetydelig. Variografisk analyse af NH4-N-inkrementer indikerer en 12-lag cyklisk variation, tolket som en mulig pseudo-variation forårsaget af systematiske analyseforsinkelser.

This thesis examines two practical measurement approaches that support quality control in small-scale production and utilities. Part 1 shows that handheld near-infrared spectroscopy (NIR)—a light-based method for estimating chemical contents—can provide fast and accurate measurements of key components in three local, low-alcohol berry and fruit wines produced in Tromsø. Two Phazir sensor systems (short- and long-wave NIR) were used with chemometric multivariate calibration. For each wine, partial least squares (PLS) regression models were built and optimized for selected winemaking components and validated using independent test sets. In the short-wave NIR region (890–1690 nm), models were satisfactory for sucrose, glucose, fructose, glycerol, ethanol, and citric, tartaric, malic, succinic, lactic, and acetic acids, and excellent for total sugar and total acid. First-derivative preprocessing was applied to all parameters except ethanol, which used full multiplicative scatter correction (MSC). These results are likely transferable to table wines and fermentation broths. Part 2 applies variographic analysis (Theory of Sampling) to assess process variability and sampling uncertainty in wastewater at the Bramming Waste Water Treatment Facility North. Both short-term incremental samples over 48 hours from two locations and 24-hour composite samples were analyzed for total phosphorus (TP), conductivity, and ammonium (NH4-N). Variograms for the increments showed a cyclic variation of approximately 420 m3, consistent with day-to-night load changes. The 24-hour composite samples were deemed reliable for emission monitoring. Over 39 days, standard composite samples (24-hour intervals) were collected from seven alternative outlet locations and compared. Variographic analysis revealed a clear 7-day cycle for TP and a 9-day cycle for conductivity, reflecting week vs. weekend patterns; the 9-day cycle remains unexplained. The standard sampling point F, directly after the weir, is representative, and point B was found to be an equally good alternative. Total Sampling Error (TSE) was estimated at 12.3% for TP and 12.4% for conductivity, while Total Analytical Error (TAE) was insignificant. Variographic analysis of NH4-N increments indicates a 12-lag cyclic variation, interpreted as a possible pseudo-variation due to systematic analysis delays.

[This abstract was generated with the help of AI]

Download
View record in AAU Student Projects