China 
Africa 
Explore chapters and articles related to this topic
Filtration
Published in Reid A. Peterson, Engineering Separations Unit Operations for Nuclear Processing, 2019
As a final note, the increase in ϕb during a dewatering operation conducted at constant pressure would manifest as a decrease in J with time. Decreasing flux with time is also observed when feeding at constant ϕb (or approximately so). Thus, use of various flux performance models for crossflow filtration need to consider the rate of change (dϕb/dt) in solids concentration fed to the filter, which impacts the use of a predictive model. The same would also be true of constant flux operations. Either way, crossflow filtration models are most useful for making qualitative judgments about performance. As of this writing, a priori quantitative performance prediction has proven to be difficult because even the simpler models with the fewest assumptions are mathematically complex and often solved numerically.
Room energy demand and thermal comfort predictions in early stages of design based on the Machine Learning methods
Published in Intelligent Buildings International, 2023
Nima Forouzandeh, Zahra Sadat Zomorodian, Zohreh Shaghaghian, Mohamad Tahsildoost
Towards these goals, the current study aims to find the most suitable ML models for energy demand and annual thermal comfort target indices by a comparative analysis between different ML approaches. Therefore, besides the common methods in literature (ANN, MR, SVM) the RF, Boosting, and ERT are utilized. Moreover, results are compared to the common physics-based (EnergyPlus) models, in terms of accuracy and calculation speeds. Results are used to develop an algorithmic framework for performance prediction in early design phases.
Optimal Empirical-Markovian approach for assessment of potential pavement rehabilitation strategies at the project level
Published in Road Materials and Pavement Design, 2018
Pavement performance prediction is a key element in pavement management modelling which is vitally needed to forecast the pavement future conditions. There are typically two general types of performance prediction models, namely deterministic and probabilistic with the latter being the most widely used one. Several versions of the probabilistic model were used by different researchers to predict pavement performance with the most popular are the Markovian-based ones (Abaza, 2016; Butt, Shahin, Carpenter, & Carnahan, 1994; Durango & Madanat, 2002; Hong & Wang, 2003; Lethanh & Adey, 2013; Mandiartha, Duffield, Thompson, & Wigan, 2012; Meidani & Ghanem, 2015). Different types of Markov chain were used by several researchers including discrete-time Markov chain, discrete-time semi-Markov chain, exponential hidden Markov chain, Poisson hidden Markov chain, random Markov chain, and recurrent Markov chain (Abaza, 2015; Lethanh & Adey, 2013; Lethanh, Kaito, & Kobayashi, 2014; Meidani & Ghanem, 2015; Yang, Lu, Gunaratne, & Dietrich, 2006; Zhang & Gao, 2012). While these researchers applied different forms of the Markov model, they all reported a good degree of success in mainly predicting the performance of original pavements. In particular, the discrete-time Markov chain with heterogeneous transition probabilities has proven to be effective in predicting the performance of original pavement (Abaza, 2015). However, limited work has been done to predict the performance of rehabilitated pavement. A main advantage of the proposed Empirical-Markovian approach is its ability to predict the heterogeneous transition probabilities for rehabilitated pavement from the corresponding ones associated with original pavement. Another advantage is its efficacy in incorporating the expected performance of potential rehabilitation strategies to become an integrated part of the prediction decision-making process.