crude processing and blending models

@article{gueddar2011novel,
  title={Novel model reduction techniques for refinery-wide energy optimisation},
  author={Taoufiq Gueddar and Vivek Dua},
  journal={Applied Energy},
  volume={89},
  number={1},
  pages={117--126},
  year={2012},
  publisher={Elsevier}
}


1 selection of the crudes
2 rigorous simulation
3 Data generation and scaling
4 NLP network training
minimize the SSE (sum of squared errors)
5 MINLP node reduction: minimize the number of nodes, while keeping the training error under the chosen tolerable error for step 4.
6 MINLP interconnection reduction: minimize the sum of the interconnection binary variables from the inputs to the hidden layer nodes and from the hidden layer to the outputs.
7 NLP training with the optimized structure
8 steps 6 and 7 carried out in a loop to choose the best run.

Fig. 2. LP techniques used in the industry to model the CDU.
Clear illustration.



@article{robertson2011multi,
  title={A multi-level simulation approach for the crude oil loading/unloading scheduling problem},
  author={Robertson, G. and Palazoglu, A. and Romagnoli, JA},
  journal={Computers and Chemical Engineering},
  volume={35},
  number={5},
  pages={817--827},
  year={2011},
  publisher={Elsevier}
}