This thesis addresses spatial interpolation and temporal prediction using air pollution data by several space-time modelling approaches. Firstly, we implement the dynamic linear modelling (DLM) approach in spatial interpolation and find various potential problems with that approach. We develop software to implement our approach. Secondly, we implement a Bayesian spatial prediction (BSP) approach to model spatio-temporal ground-level ozone fields and compare the accuracy of that approach with that of the DLM. Thirdly, we develop a Bayesian version empirical orthogonal function (EOF) method to incorporate the uncertainties due to temporally varying spatial process, and the spatial variations at broad- and fine- scale. Finally, we extend the BSP into the DLM framework to develop a unified Bayesian spatio-temporal model for univariate and multivariate responses. The result generalizes a number of current approaches in this field.

This dissertation presents a multilevel model of national-level factors and their impact on the organizational-level characteristics of social enterprises and their innovations. This study builds on the foundations of two theoretical frameworks: the national systems of innovation, which recognizes economic competitiveness to be a product of several interrelated institutions (e.g. financial, educational, cultural, historical) and where organizational-level innovation drives country level competitiveness; and the comparative social enterprise framework, which contends that national-level institutions (e.g., economic competitiveness, models of civil society) drive the size and shape of the social enterprise sector of a country. Data for this study were collected from multiple secondary global datasets representing 54 countries across seven world regions. Research questions and hypotheses are examined using ordinal and logistic hierarchical generalized linear modeling, two analytical techniques capable of explaining variation at one level (i.e., organizations) as a consequence of factors at another level of analysis (i.e., countries) for non-normally distributed dependent variables. Findings indicate that economic competitiveness, welfare spending, culture and quality of life significantly impact the odds of a business being a social enterprise. Fewer significant relationships were found social enterprise innovations. Conclusions and policy implications are discussed in light of data limitations and the current state of the field.

This thesis addresses spatial interpolation and temporal prediction using air pollution data by several space-time modelling approaches. Firstly, we implement the dynamic linear modelling (DLM) approach in spatial interpolation and find various potential problems with that approach. We develop software to implement our approach. Secondly, we implement a Bayesian spatial prediction (BSP) approach to model spatio-temporal ground-level ozone fields and compare the accuracy of that approach with that of the DLM. Thirdly, we develop a Bayesian version empirical orthogonal function (EOF) method to incorporate the uncertainties due to temporally varying spatial process, and the spatial variations at broad- and fine- scale. Finally, we extend the BSP into the DLM framework to develop a unified Bayesian spatio-temporal model for univariate and multivariate responses. The result generalizes a number of current approaches in this field.

This thesis addresses spatial interpolation and temporal prediction using air pollution data by several space-time modelling approaches. Firstly, we implement the dynamic linear modelling (DLM) approach in spatial interpolation and find various potential problems with that approach. We develop software to implement our approach. Secondly, we implement a Bayesian spatial prediction (BSP) approach to model spatio-temporal ground-level ozone fields and compare the accuracy of that approach with that of the DLM. Thirdly, we develop a Bayesian version empirical orthogonal function (EOF) method to incorporate the uncertainties due to temporally varying spatial process, and the spatial variations at broad- and fine- scale. Finally, we extend the BSP into the DLM framework to develop a unified Bayesian spatio-temporal model for univariate and multivariate responses. The result generalizes a number of current approaches in this field. / Science, Faculty of / Statistics, Department of / Graduate

Abstract Ecological research based on both species and their traits help us to understand the main mechanisms and environmental factors structuring biological communities. In general, variation in community composition is thought to be a consequence of both stochastic and deterministic factors. In stream ecology, the traditional view has been that the local habitat conditions pose a strong environmental filter that selects only species with the right functional traits into the local communities. However, recent studies on streams have also suggested that the responses of species to environmental gradients may be independent of those of other species due to stochastic factors, such as species dispersal, which then result in more continuous communities along environmental gradients. The aim of this thesis was to explore the relative importance of the deterministic and stochastic factors in the structuring of taxonomic and functional trait-based macroinvertebrate communities in streams in a high-latitude catchment by comparing the variation in these community facets along environmental and spatial gradients. Also, the relationship between environment and the functionally-defined communities was explored closely. The results indicated how the taxonomic composition of the communities may be more closely related to the stochastic and dispersal-related factors, whereas the functional composition of the communities may be more closely related to the deterministic environmental filtering processes. However, the overall structure of the communities seems to be strongly controlled by the variation in environment, although the heterogeneous and harsh conditions of the streams may preclude the formation of predictable community types. Nonetheless, some noticeable responses of different traits to different environmental factors were found, suggesting that definable functional trait-environment relationships may be discovered if key traits of the species can be identified. Overall, these findings underline the benefits of describing both taxonomic and functional-based communities when exploring the mechanisms behind the structuring of macroinvertebrate communities. The results also have applications for conservation practices. Conservation efforts should focus on varying environmental conditions in order to cover all aspects of macroinvertebrate community variation. / Tiivistelmä Lajeihin ja lajien toiminnallisiin lajiominaisuuksiin pohjautuva ekologinen tutkimus tuo uutta tietoa biologisten yhteisöjen taustalla vaikuttavista tekijöistä. Yleisesti yhteisöjen rakentumiseen vaikuttavat niin deterministiset kuin stokastiset ympäristössä vaikuttavat tekijät. Virtavesiyhteisöjen on perinteisesti ajateltu rakentuneen niin sanottujen ympäristösuodattimien mukaisesti, jolloin ympäristön vaihtelu suodattaa tietynlaisiin ympäristöihin vain lajit, joilla on tarvittavat ominaisuudet paikalla selviytyäkseen. Useat viimeaikaiset tutkimukset ovat kuitenkin osoittaneet virtavesiyhteisöissä elävien lajien esiintymisen vaihtelevan ympäristössä myös itsenäisesti erilaisista stokastisista, kuten lajien dispersaaliin vaikuttavista, tekijöistä johtuen. Tässä väitöstutkimuksessa tutkin näiden determinististen ja stokastisten ympäristötekijöiden suhteellisia vaikutuksia taksonomisesti ja toiminnallisesti luokiteltujen pohjaeläinyhteisöjen rakentumiseen pohjoisissa virtavesissä. Myös yksittäisten lajiominaisuuksien ja toiminnallisten yhteisöjen suhde pohjoisten virtavesien ympäristöolosuhteisiin oli tarkastelun alla. Tutkimuksen tulokset antoivat viitteitä siitä, että ympäristössä toimivat stokastiset ja lajien dispersaaliin liittyvät tekijät vaikuttaisivat voimakkaammin taksonomisesti luokiteltujen yhteisöjen vaihteluun, kun taas toiminnallisesti luokitellut yhteisöt vaikuttaisivat rakentuneen enemmän determinististen ympäristöprosessien mukaisesti. Kokonaisuudessaan yhteisöt vaikuttaisivat kuitenkin rakentuneen voimakkaasti vaihtelevien ympäristöolosuhteiden ohjaamana, ja tämä vaihtelu voi estää selkeästi ennustettavien yhteisörakenteiden synnyn. Muutamia selkeitä lajiominaisuusvasteita kuitenkin löytyi, mikä antaa viitteitä ennustettavissa olevien toiminnallisten yhteisöjen olemassaolosta, mikäli yhteisöjen menestymisen kannalta merkittävimmät lajiominaisuudet vain osataan määrittää. Nämä tulokset osoittavat, miten sekä taksonomisesti että toiminnallisesti luokiteltujen yhteisöjen käyttäminen rinnakkain yhteisöekologisissa tutkimuksissa voi auttaa selventämään yhteisöjen synnyn taustalla vaikuttavia tekijöitä. Tuloksilla on merkitystä myös virtavesiyhteisöjen suojelun kannalta. Suojelutoimenpiteet tulisi kohdistaa kattamaan ympäristöolosuhteita laajasti, jotta ympäristöolosuhteiden mukaan vaihtelevat yhteisöt tulisivat parhaalla mahdollisella tavalla katetuiksi.

The frequency of the Nordic power system has been increasingly deviating outside the normal frequency band (50 ± 0.1 Hz) in the past two decades. In an effort to counteract this, the Nordic Transmission System Operators (TSOs) have proposed new and stricter requirements on the units participating in Frequency Containment Reserve (FCR) market. The implication of these new requirements is that all units participating in the FCR markets must be re-evaluated and early predictions state that double regulated (DR) hydropower turbines, such as Kaplan and Bulb types, will have problems meeting these new requirements. To limit the amount of work required to re-evaluate all DR turbines, which might ultimately show that the unit will not pass the requirements, it is highly beneficial for power producing companies to be able to simulate the turbine response accurately enough, without full-scale physical testing, to filter out the possible cases from the impossible ones. This thesis is aimed at finding a hydropower turbine model that could accurately predict the FCR prequalification outcome for a DR hydropower turbine in Simulink from only pre-existing testing and design data. Three hydropower turbine models, one single regulated (SR) and two DR, were tested and validated for FCR Normal operation (FCR-N). The validation was performed by comparing the model results to field test data attained from a full-scale FCR-N test performed on one of Fortum’s DR hydropower units. Furthermore, two of these models were simulated and tested with regard to the FCR Disturbance (FCR-D) requirements but not validated as no such data was available. The results showed that simulating DR turbines with SR models can give inaccurate results as the SR model could not capture the more complex dynamics of the DR turbine accurately enough. The SR model underestimated the active power response drastically while at the same time overestimating the dynamic stability. The DR turbine models managed to capture the dynamics better with the most prominent model, DR Model 2, giving less than 3 % overestimation of the FCR-N capacity and a correctly predicting that the unit would fail the dynamic stability for FCR-N. The thesis results further showed that when modelling DR turbines, the calculation of the net head is vital for accurate model response. Finally, both the SR and DR model agreed that the unit would likely not pass the FCR-D requirements and therefore confirm the early predictions that DR hydropower turbines, such as the one modelled in this thesis, may have problems passing the new FCR prequalification requirements. Ultimately this might impact electrical price or frequency quality negatively. / Frekvenskvaliteten i det nordiska kraftsystemet har stadigt avvikit alltmer från det normala frekvensbandet (50 ± 0,1 Hz) under de två senaste decennierna. I ett försöka att motverka denna trend har de nordiska kraftsystemoperatörerna (TSO) föreslagit nya och striktare krav på aggregat som deltar i frekvensreservsmarknaden (FCR). Dessa nya krav innebär att alla aggregat som redan deltar på FCR-marknaden måste omvärderas och de bedömningar som har gjorts antyder att dubbelreglerade (DR) vattenkraftturbiner, som kaplan- och bulbturbiner, kommer att ha svårt med att uppfylla dessa nya krav. För att begränsa mängden arbete som krävs för att omvärdera alla DR-turbiner, som i slutändan kan visa att aggregatet inte kommer att klara kraven, är det mycket för fördelaktigt för kraftföretag att kunna noggrant simulera turbinresponsen innan de fysiska testerna utförs, för att filtrera bort de omöjliga fallen från de möjliga. Detta examensarbetes syfte var att hitta en turbinmodell som med bra noggrannhet kan förutsäga FCR förkvalificeringsresultatet for en dubbelreglerad vattenkraftturbin, med hjälp av Simulink från endast befintliga driftdata och information om aggregatets konstruktion. Tre turbinmodeller, en enkelreglerade (SR) och två dubbelreglerade, testades och validerades för FCR Normal drift (FCR-N). Valideringen utfördes genom att jämföra modellresultaten med fälttestdata från ett fullskaligt FCR-N-test som utfördes på en av Fortums dubbelreglerade vattenkraftturbiner. Dessutom simulerades även två av dessa modeller och testades med avseende på frekvensreglering vid störd drift (FCR-D) men validerades inte på grund av avsaknad av fälttestdata. Resultaten visade att simulering av dubbelreglerade turbiner med enkelreglerade turbinmodeller kan vara problematisk eftersom enkelreglerade modellen inte kunde fånga dubbelregleringsturbinernas mer komplexa dynamik tillräckligt bra. Enkelreglerade modellerna tydligt underskattade den aktiva effektresponsen samtidigt som de överskattade den dynamiska stabiliteten. De dubbelreglerade turbinmodellerna fångade däremot dynamiken bättre, där den mest framstående modellen, DR model 2, gav mindre än 3 % överskattning av FCR-N-kapaciteten och en korrekt förutsägelse att enheten skulle misslyckas uppfylla den dynamiska stabiliteten för FCR-N. Resultaten visade vidare att beräkningen av nettofallhöjden är avgörande för korrekt modellrespons. Slutligen var både singel- och dubbelreglerade modellerna överens om att aggregatet sannolikt inte skulle klara FCR-D-kraven och bekräftar därmed de tidiga bedömningarna att dubbelreglerade aggregat, som den som modellerades i denna avhandling, kan ha svårt att klara de nya FCR kraven. I slutändan kan det påverka såväl elpris som frekvenskvalitet negativt.

Tesis por compendio / The Lack of Homogeneity in the Product (LHP) appears in productive processes with raw materials, which directly stem from nature and/or production processes with operations that confer heterogeneity to the characteristics of the outputs obtained, even when the inputs used are homogeneous. LHP appears in different sectors such as ceramic tile, horticulture, marble, snacks, among others. LHP becomes a managerial problem when customers require to be served with homogeneous product. Supply chains responsible to provide homogeneous product face the need to include classification activities in their productive processes to obtain sub-lots of homogeneous product. Due to the inherent LHP uncertainty, these homogeneous sub-lots will not be known until the product have been produced and classified. An improper management of the LHP can have a very negative impact on the customers' satisfaction due to inconsistencies in the answer to their requirements and also on the Supply Chain's efficiency. The Order Promising Process (OPP) appears as a key element for properly managing the LHP in order to ensure the matching of uncertain homogeneous supply with customer order proposals. The OPP refers to the set of business activities that are triggered to provide a response to the orders from customers. These activities are related to the acceptance/rejection decision, and to set delivery dates. For supply chains affected by the LHP, the OPP must consider the homogeneity as another requirement in the answer to the orders. Besides, due to the LHP inherent uncertainty, discrepancies between the real and planned homogeneous quantities might provoke that previously committed orders cannot be served. The Shortage Planning (SP) process intends to find alternatives in order to minimise the negative impact on customers and the supply chain. Considering LHP in the OPP brings a set of new challenging features to be addressed. The conventional approach of assuming homogeneity in the product for the master production schedule (MPS) and the quantities Available-To-Promise (ATP) derived from it is no longer adequate. Instead, both the MPS and ATP should be handled in terms of homogeneous sub-lots. Since the exact quantity of homogeneous product from the planned lots in the MPS is not exactly known until the classification activities have been performed, the ATP also inherits this uncertainty, bringing a new level of complexity. Non-homogeneous product cannot be accumulated in order to fulfil future incoming orders. Even more, if the product handled is perishable, the homogeneity management becomes considerably more complex. This is because the state of the product is dynamic with time and related variables to it, like quality, price, etc., could change with time. This situation could bring unexpected wasting costs apart from the shortages already mentioned. The perishability factor is itself another source of uncertainty associated to the LHP. This dissertation proposes a conceptual framework and different mathematical programming models and tools, in both deterministic and uncertainty environments, in order to support the OPP and SP under LHP's effect. The aim is to provide a reliable commitment with customer orders looking for a high service level not just in the due date and quantity but also in the homogeneity requirements. The modelling of the characteristics inherent to LHP under deterministic context constitutes itself one of the main contribution of this dissertation. Another novelty consists in the inclusion of uncertainty in the definition of homogeneous sub-lots, their quantities and their dynamic state and value. The uncertainty modelling approach proposed is mainly based on the application of fuzzy set theory and possibility theory. The proposed mathematical models and tools have been validated in real cases of SC, specifically in the ceramic tile sector for non perishables, and in the fruit sector for perishables. The results show a ... / La Falta de Homogeneidad en el Producto (LHP, por sus siglas del inglés ``Lack of Homogeneity in the Product'') aparece en procesos productivos con materias primas que derivan directamente de la naturaleza y/o procesos de producción con operaciones que confieren heterogeneidad a las características de los productos obtenidos, incluso cuando los insumos utilizados son homogéneos. La LHP aparece en diferentes sectores como la cerámica, horticultura, mármol, snacks, entre otros. Se convierte en un problema gerencial cuando los clientes requieren homogeneidad en el producto y las cadenas de suministro enfrentan la necesidad de incluir actividades de clasificación en sus procesos productivos para obtener sub-lotes de producto homogéneo. Debido a la incertidumbre inherente a la LHP, los sub-lotes homogéneos y su cantidad no serán conocidos hasta que el producto haya sido producido y clasificado. Una gestión inadecuada de la LHP puede tener un impacto muy negativo en la satisfacción de los clientes debido a inconsistencias en la respuesta a sus requerimientos y también en la eficacia de la Cadena de Suministro. El Proceso de Comprometer de Pedido (OPP, por sus siglas del inglés ``Order Promising Process'') aparece como un elemento clave para gestionar adecuadamente la LHP, con el fin de asegurar la coincidencia entre el suministro incierto de producto homogéneo y las propuestas de pedido del cliente. El OPP se refiere al conjunto de actividades empresariales realizadas para proporcionar una respuesta a las órdenes de los clientes. Estas actividades están relacionadas con las decisiones de aceptación/rechazo, y establecimiento de fechas de entrega para las órdenes del cliente. En las cadenas de suministro afectadas por la LHP, el OPP debe considerar la homogeneidad como otro requisito adicional en la respuesta a los pedidos. Además, debido a la incertidumbre intrínseca de la LHP, las discrepancias entre las cantidades homogéneas reales y planificadas podrían provocar que las órdenes comprometidas anteriormente no puedan ser completadas debido a la escasez de producto. El proceso de planificación de la escasez (SP, por sus siglas del inglés "Shortage Planning") se encarga de encontrar alternativas para minimizar este impacto negativo en los clientes y la cadena de suministro. Considerar la LHP dentro del OPP implica un conjunto nuevo de características desafiantes que deben ser abordadas. El enfoque convencional de asumir la homogeneidad en el producto para el programa maestro de producción (MPS, por sus siglas del inglés "Master Production Schedule") y las cantidades disponibles a comprometer (ATP, por sus siglas del inglés "Available-To-Promise") derivadas de él, no es adecuado. En cambio, tanto el MPS como el ATP deben manejarse en términos de sub-lotes homogéneos. Dado que la cantidad exacta de producto homogéneo de los lotes previstos en el MPS no se sabe exactamente hasta que se han realizado las actividades de clasificación, el ATP también hereda esta incertidumbre, trayendo un nuevo nivel de complejidad. El producto no homogéneo no se puede acumular para satisfacer futuras órdenes entrantes. Más aún, si el producto manipulado es perecedero, el manejo de la homogeneidad se vuelve mucho más complejo. Esto se debe a que el estado del producto es dinámico en el tiempo, y variables relacionadas como calidad, precio, etc., podrían también cambiar con el tiempo. Esta situación puede provocar costos inesperados de desperdicio aparte de la escasez ya mencionada. El factor de perecedero es en sí mismo otra fuente de incertidumbre asociada a la LHP. Esta disertación propone un marco conceptual y diferentes modelos y herramientas de programación matemática, tanto en entornos deterministas como de incertidumbre, para apoyar al OPP y SP considerando el efecto de LHP. El objetivo es proporcionar un compromiso fiable con los pedidos de los clientes en busca de un alto nivel de servicio no s / La Falta d'Homogeneïtat en el Producte (LHP, per les seues sigles de l'anglés ''Lack of Homogeneity in the Product'') apareix en processos productius amb matèries primes que deriven directament de la natura i/o processos de producció amb operacions que conferixen heterogeneïtat a les característiques dels productes obtinguts, fins i tot quan les entrades utilitzades són homogènies . La LHP apareix en diferents sectors com la ceràmica, horticultura, marbre, snacks, entre altres. Es convertix en un problema gerencial quan els clients requereixen homogeneïtat en el producte i les cadenes de subministrament enfronten la necessitat d'incloure activitats de classificació en els seus processos productius per a obtindre sublots de producte homogeni. A causa de la incertesa inherent a la LHP, els sublots homogenis i la seua quantitat no seran coneguts fins que el producte haja sigut produït i classificat. Una gestió inadequada de la LHP pot tindre un impacte molt negatiu en la satisfacció dels clients degut a inconsistències en la resposta als seus requeriments i també en l'eficàcia de la Cadena de Subministrament. El Procés de Comprometre Comandes (OPP, per les seues sigles de l'anglés ''Order Promising Process'') apareix com un element clau per a gestionar adequadament la LHP, a fi d'assegurar la coincidència entre el subministrament incert de producte homogeni i les propostes de comanda del client. L'OPP es refereix al conjunt d'activitats empresarials realitzades per a proporcionar una resposta a les ordres dels clients. Aquestes activitats estan relacionades amb les decisions d'acceptació/rebuig, i establiment de dates de lliurament per a les ordres del client. En les cadenes de subministrament afectades per la LHP, l'OPP ha de considerar l'homogeneïtat com un altre requisit addicional en la resposta a les comandes. A més, a causa de la incertesa intrínseca de la LHP, les discrepàncies entre les quantitats homogènies reals i planificades podrien provocar que les ordres compromeses anteriorment no puguen ser completades a causa de l'escassetat de producte. El procés de planificació de l'escassetat (SP, per les seues sigles de l'anglés "Shortage Planning") s'encarrega de trobar alternatives per a minimitzar aquest impacte negatiu en els clients i en la cadena de subministrament. Considerar la LHP dins de l'OPP implica un conjunt nou de característiques desafiants que han de ser abordades. L'enfocament convencional d'assumir l'homogeneïtat en el producte per al programa mestre de producció (MPS, per les seues sigles de l'anglés "Master Production Schedule") i les quantitats disponibles a comprometre (ATP, per les seues sigles de l'anglés "Available-To-Promise") derivades d'ell, no és adequat. En canvi, tant el MPS com l'ATP han de manejar-se en termes de sublots homogenis. Atés que la quantitat exacta de producte homogeni dels lots previstos en el MPS no se sap exactament fins que s'han realitzat les activitats de classificació, l'ATP també hereta aquesta incertesa, portant un nou nivell de complexitat. El producte no homogeni no es pot acumular per a satisfer futures ordees entrants. Més encara, si el producte manipulat és perible, el maneig de l'homogeneïtat es torna molt més complex. Açò es deu al fet que l'estat del producte és dinàmic en el temps, i variables relacionades com qualitat, preu, etc., podrien també canviar amb el temps. Aquesta situació pot provocar costos inesperats de rebuig a banda de l'escassetat ja esmentada. El factor de perible és en si mateix un altra font d'incertesa associada a la LHP. Aquesta dissertació proposa un marc conceptual i diferents models i eines de programació matemàtica, tant en entorns deterministes com d'incertesa, per a recolzar a l'OPP i SP considerant l'efecte de LHP. L'objectiu és proporcionar un compromís fiable amb les comandes dels clients a la recerca d'un alt nivell de servei no sols en la data i la quantitat esperades, s / Grillo Espinoza, H. (2017). Advanced methods and models in uncertainty for the order promising process in supply chain characterized by the lack of homogeneity in product [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/91110 / Compendio