Simulation of greenhouse gases following land-use change to bioenergy crops using the ECOSSE model : a comparison between site measurements and model predictions

This work contributes to the ELUM (Ecosystem Land Use Modelling & Soil Carbon GHG Flux Trial) project, which was commissioned and funded by the Energy Technologies Institute (ETI). We acknowledge the E-OBS data set from the EU-FP6 project ENSEMBLES (http://ensembles-eu.metoffice.com) and the data providers in the ECA&D project (http://www.ecad.eu).

An evaluation of physician predictions of discharge on a general medicine service.

The goal of this study was to evaluate general medicine physicians' ability to predict hospital discharge. We prospectively asked study subjects to predict whether each patient under their care would be discharged on the next day, on the same day, or neither. Discharge predictions were recorded at 3 time points: mornings (7-9 am), midday (12-2 pm), or afternoons (5-7 pm), for a total of 2641 predictions. For predictions of next-day discharge, the sensitivity (SN) and positive predictive value (PPV) were lowest in the morning (27% and 33%, respectively), but increased by the afternoon (SN 67%, PPV 69%). Similarly, for same-day discharge predictions, SN and PPV were highest at midday (88% and 79%, respectively). We found that although physicians have difficulty predicting next-day discharges in the morning prior to the day of expected discharge, their ability to correctly predict discharges continually improved as the time to actual discharge decreased. Journal of Hospital Medicine 2015;10:808-810. © 2015 Society of Hospital Medicine.

Assessment of turbulence model predictions for a centrifugal compressor simulation

Steady-state computational fluid dynamics (CFD) simulations are an essential tool in the design process of centrifugal compressors. Whilst global parameters, such as pressure ratio and efficiency, can be predicted with reasonable accuracy, the accurate prediction of detailed compressor flow fields is a much more significant challenge. Much of the inaccuracy is associated with the incorrect selection of turbulence model. The need for a quick turnaround in simulations during the design optimisation process, also demands that the turbulence model selected be robust and numerically stable with short simulation times.
In order to assess the accuracy of a number of turbulence model predictions, the current study used an exemplar open CFD test case, the centrifugal compressor ‘Radiver’, to compare the results of three eddy viscosity models and two Reynolds stress type models. The turbulence models investigated in this study were (i) Spalart-Allmaras (SA) model, (ii) the Shear Stress Transport (SST) model, (iii) a modification to the SST model denoted the SST-curvature correction (SST-CC), (iv) Reynolds stress model of Speziale, Sarkar and Gatski (RSM-SSG), and (v) the turbulence frequency formulated Reynolds stress model (RSM-ω). Each was found to be in good agreement with the experiments (below 2% discrepancy), with respect to total-to-total parameters at three different operating conditions. However, for the off-design conditions, local flow field differences were observed between the models, with the SA model showing particularly poor prediction of local flow structures. The SST-CC showed better prediction of curved rotating flows in the impeller. The RSM-ω was better for the wake and separated flow in the diffuser. The SST model showed reasonably stable, robust and time efficient capability to predict global and local flow features.

Comparison of DES predictions of turbulent flow and heat transfer in a 45° ribbed duct

Numerical predictions of the turbulent flow and heat transfer of a stationary duct with square ribs 45° angled to the main flow direction are presented. The rib height to channel hydraulic diameter is 0.1, the rib pitch to rib height is 10. The calculations have been carried out for a bulk Reynolds number of 50,000. The flows generated by ribs are dominated by separating and reattaching shear layers with vortex shedding and secondary flows in the cross-section. The hybrid RANS-LES approach is adopted to simulate such flows at a reasonable computation cost. The capability of the various versions of DES method, depending the RANS model, such as DES-SA, DES-RKE, DES-SST, have been compared and validated against the experiment. The significant effect of RANS model on the accuracy of the DES prediction has been shown. The DES-SST method, which was able to reproduce the correct physics of flow and heat transfer in a ribbed duct showed better performance than others.

A comparison of predictions from the buildingEXODUS evacuation model with experimental data

Abstract not available

Thermal Tolerance of the Coffee Berry Borer Hypothenemus hampei: Predictions of Climate Change Impact on a Tropical Insect Pest

Coffee is predicted to be severely affected by climate change. We determined the thermal tolerance of the coffee berry borer, Hypothenemus hampei, the most devastating pest of coffee worldwide, and make inferences on the possible effects of climate change using climatic data from Colombia, Kenya, Tanzania, and Ethiopia. For this, the effect of eight temperature regimes (15, 20, 23, 25, 27, 30, 33 and 35 degrees C) on the bionomics of H. hampei was studied. Successful egg to adult development occurred between 20-30 degrees C. Using linear regression and a modified Logan model, the lower and upper thresholds for development were estimated at 14.9 and 32 degrees C, respectively. In Kenya and Colombia, the number of pest generations per year was considerably and positively correlated with the warming tolerance. Analysing 32 years of climatic data from Jimma (Ethiopia) revealed that before 1984 it was too cold for H. hampei to complete even one generation per year, but thereafter, because of rising temperatures in the area, 1-2 generations per year/coffee season could be completed. Calculated data on warming tolerance and thermal safety margins of H. hampei for the three East African locations showed considerably high variability compared to the Colombian site. The model indicates that for every 1 degrees C rise in thermal optimum (T(opt)), the maximum intrinsic rate of increase (r(max)) will increase by an average of 8.5%. The effects of climate change on the further range of H. hampei distribution and possible adaption strategies are discussed. Abstracts in Spanish and French are provided as supplementary material Abstract S1 and Abstract S2.

Validation of PigBal model predictions for pig manure production

PigBal is a mass balance model that uses pig diet, digestibility and production data to predict the manure solids and nutrients produced by pig herds. It has been widely used for designing piggery effluent treatment systems and sustainable reuse areas at Australian piggeries. More recently, PigBal has also been used to estimate piggery volatile solids production for assessing greenhouse gas emissions for statutory reporting purposes by government, and for evaluating the energy potential from anaerobic digestion of pig effluent. This paper has compared PigBal predictions of manure total, volatile, and fixed solids, and nitrogen (N), phosphorus (P) and potassium (K), with manure production data generated in a replicated trial, which involved collecting manure from pigs housed in metabolic pens. Predictions of total, volatile, and fixed solids and K in the excreted manure were relatively good (combined diet R2 ≥ 0.79, modelling efficiency (EF) ≥ 0.70) whereas predictions of N and P, were generally less accurate (combined diet R2 0.56 and 0.66, EF 0.19 and –0.22, respectively). PigBal generally under-predicted lower N values while over-predicting higher values, and generally over-predicted manure P production for all diets. The most likely causes for this less accurate performance were ammonium-N volatilisation losses between manure excretion and sample analysis, and the inability of PigBal to account for higher rates of P uptake by pigs fed diets containing phytase. The outcomes of this research suggest that there is a need for further investigation and model development to enhance PigBal’s capabilities for more accurately assessing nutrient loads. However, PigBal’s satisfactory performance in predicting solids excretion demonstrates that it is suitable for assessing the methane component of greenhouse gas emission and the energy potential from anaerobic digestion of volatile solids in piggery effluent. The apparent overestimation of N and P excretion may result in conservative nutrient application rates to land and the over-prediction of the nitrous oxide component of greenhouse gas emissions.

Segregation predictions for a binary mixture of granular material under flow conditions in a continuum mechanics framework

Abstract not available

Validation of PigBal model predictions for pig manure production

PigBal is a mass balance model that uses pig diet, digestibility and production data to predict the manure solids and nutrients produced by pig herds. It has been widely used for designing piggery effluent treatment systems and sustainable reuse areas at Australian piggeries. More recently, PigBal has also been used to estimate piggery volatile solids production for assessing greenhouse gas emissions for statutory reporting purposes by government, and for evaluating the energy potential from anaerobic digestion of pig effluent. This paper has compared PigBal predictions of manure total, volatile, and fixed solids, and nitrogen (N), phosphorus (P) and potassium (K), with manure production data generated in a replicated trial, which involved collecting manure from pigs housed in metabolic pens. Predictions of total, volatile, and fixed solids and K in the excreted manure were relatively good (combined diet R2 ≥ 0.79, modelling efficiency (EF) ≥ 0.70) whereas predictions of N and P, were generally less accurate (combined diet R2 0.56 and 0.66, EF 0.19 and -0.22, respectively). PigBal generally under-predicted lower N values while over-predicting higher values, and generally over-predicted manure P production for all diets. The most likely causes for this less accurate performance were ammonium-N volatilisation losses between manure excretion and sample analysis, and the inability of PigBal to account for higher rates of P uptake by pigs fed diets containing phytase. The outcomes of this research suggest that there is a need for further investigation and model development to enhance PigBal's capabilities for more accurately assessing nutrient loads. However, PigBal's satisfactory performance in predicting solids excretion demonstrates that it is suitable for assessing the methane component of greenhouse gas emission and the energy potential from anaerobic digestion of volatile solids in piggery effluent. The apparent overestimation of N and P excretion may result in conservative nutrient application rates to land and the over-prediction of the nitrous oxide component of greenhouse gas emissions. © CSIRO 2016.

Improvement of Virtual Screening Predictions using Computational Intelligence Methods

Virtual Screening (VS) methods can considerably aid clinical research, predicting how ligands interact with drug targets. However, the accuracy of most VS methods is constrained by limitations in the scoring function that describes biomolecular interactions, and even nowadays these uncertainties are not completely understood. In order to improve accuracy of scoring functions used in most VS methods we propose a hybrid novel approach where neural networks (NNET) and support vector machines (SVM) methods are trained with databases of known active (drugs) and inactive compounds, this information being exploited afterwards to improve VS predictions.

SEQUENTIAL DECISIONS AND PREDICTIONS IN NATURAL LANGUAGE PROCESSING

Natural language processing has achieved great success in a wide range of ap- plications, producing both commercial language services and open-source language tools. However, most methods take a static or batch approach, assuming that the model has all information it needs and makes a one-time prediction. In this disser- tation, we study dynamic problems where the input comes in a sequence instead of all at once, and the output must be produced while the input is arriving. In these problems, predictions are often made based only on partial information. We see this dynamic setting in many real-time, interactive applications. These problems usually involve a trade-off between the amount of input received (cost) and the quality of the output prediction (accuracy). Therefore, the evaluation considers both objectives (e.g., plotting a Pareto curve). Our goal is to develop a formal understanding of sequential prediction and decision-making problems in natural language processing and to propose efficient solutions. Toward this end, we present meta-algorithms that take an existent batch model and produce a dynamic model to handle sequential inputs and outputs. Webuild our framework upon theories of Markov Decision Process (MDP), which allows learning to trade off competing objectives in a principled way. The main machine learning techniques we use are from imitation learning and reinforcement learning, and we advance current techniques to tackle problems arising in our settings. We evaluate our algorithm on a variety of applications, including dependency parsing, machine translation, and question answering. We show that our approach achieves a better cost-accuracy trade-off than the batch approach and heuristic-based decision- making approaches. We first propose a general framework for cost-sensitive prediction, where dif- ferent parts of the input come at different costs. We formulate a decision-making process that selects pieces of the input sequentially, and the selection is adaptive to each instance. Our approach is evaluated on both standard classification tasks and a structured prediction task (dependency parsing). We show that it achieves similar prediction quality to methods that use all input, while inducing a much smaller cost. Next, we extend the framework to problems where the input is revealed incremen- tally in a fixed order. We study two applications: simultaneous machine translation and quiz bowl (incremental text classification). We discuss challenges in this set- ting and show that adding domain knowledge eases the decision-making problem. A central theme throughout the chapters is an MDP formulation of a challenging problem with sequential input/output and trade-off decisions, accompanied by a learning algorithm that solves the MDP.