GT China coal model

Overview

GT China coal model

The full version of a China coal transport model with a very high spatial reslution.

What it does

The code works in a few steps:

  1. Take easily understandable and readable xlsx input files on networks, plants, demand etc, and create project build files form this (done in R).
  2. Take the build files and create an LP problem file from it (in python, either locally or on AWS Sagemaker).
  3. Solve the problem from the LP problem file, and write solution to a txt file (either in Cplex interactive or in python).
  4. Process the solution txt file and write to easily understandable and readable xlsx (in R). The packages required to run these scripts are included in the environment.yml (for python) and the renv.lock file (for R; this first requires installation of renv package: https://rstudio.github.io/renv/articles/renv.html; after installation run renv:restore()).

The model

The model optimizes for a minimum cost of production + transport + transmission.
Production meaning coal mining costs, transport meaning rail/truck/riverborne/ocean-going transport and handling costs, and transmission meaning inter-provincial transport of electricity via UHV cables.

Constraints in the optimization

The constraints in the mini testbench are the same as in the full model. These are:

  • Mines (or any other node) cannot supply types of coal they do not produce.
  • The flow of coal of each type out of a node cannot exceed flows of coal of each type into a node plus supply by the node (with supply being non-zero only for mines).
  • The energy content of the supply and the flows of coal of each type into a node have to be at least equal to the demand for electricity, plus other thermal coal demand, plus the energy content of flows of coal of each type out of a node. Note that only mines can supply coal, all demand for electrical power occurs in provincial demand nodes, and demand for other thermal coal is placed at city-level nodes.
  • The amount of hard coking coal (HCC) flowing into a node has to at least be equal to the steel demand multiplied by 0.581. Note that all steel demand is placed in provincial level steel demand nodes, which are connecte with uni-directional links from steel plants to steel demand nodes. This means no coal can flow out of a steel demand node and we do not need further forumalae for mass balances. Also note that we presume a mix of coking coal need to produce a ton of steel of 581 kg Hard coking coal (HCC), 176 kg of soft coking coal (SCC), and 179 kg of pulverized coal for injection (PCI).
  • The amount of soft coking coal (SCC) flowing into a node has to at least be equal to the amount of HCC flowing into that node, mulitplied with 0.581/0.176.
  • The amount of pulverized coal for injection (PCI) flowing into a node has to at least be equal to the amount of HCC flowing into that node, mulitplied with 0.581/0.179.
  • The total volume of all coal types transported along a link cannot exceed the transport capacity of that link. Note that this constraint is applied only to those links with a non-infinite transport capacity. In practive this means rail links are assumed to have a transport capacity, ocean routes, rivers, and road links are assumed to have infinite capacity.
  • The total amount of energy transported along a link cannot exceed the transmission capacity of that link. That is, the amount of each coal type multiplied with the energy content of each coal type cannot exceed the electrical transmission capacity of links. This constraint is applied only links between power plant units and provincial electricity demand nodes, as well as UHV transmission links between provincial electricity demand nodes. These are the only links along which electrical energy is transported. All other links transport physcial quantities of coal. This line simultaneously deals with the production capapcity (MW) and conversion efficiency of power plants: the energy transported over a link cannot exceed the volume of each coal type, multiplied with the energy content of each coal type, multiplied with the energy conversion factor of the link. For links between coal fired power plant units and provincial electricity demand nodes, this is equal to the conversion effincy of the power plant unit. For UHV transmission links between two provincial level electricity demand nodes, this is equal to (1- transmission losses) over that UHV line, with transmission losses calculated based on transmission distance and a benchmark loss for UHV-DC or UHV-AC lines.
  • The handling capacity of ports cannot be exceeded. Specifically, the total amount of coal flowing out of a port cannot exceed its handling capacity.
  • The production capacity of steel plants cannot be exceeded. Specifically, the total amount of hard coking coal, soft coking coal, and pulverized coal for injection flowing out of a steel plant node (and towards a provincial steel demand node) cannot exceed the steel plant's production capacity multiplied by 0.581+0.176+0.179, the mix of different coking coals needed to produce steel.

Technical notes

  • All transport costs are pre-calculated for each link, and include a fixed handling costs and a distance based transport cost, based on the type handling (origin and destaination) and type of transport (separate for rail, truck, riverborne, ocean-going. A small number of coal rail lines has specific handling and transport costs).
  • Some of the capacities are already reported in the input sheet for the edges. The physical transport capacity from this sheet is used. For capacities of ports, steel plants, and electrical transmission capacities, the data from the separate port/steel plant/electrical capacities sheets is used.
  • An exmaple lp file is included to make this reporsitory as self-conatined as possible. This lp file is zipped to stay within github file size limits.

Contributions

This project was developed by:

  • Jorrit Gosens: conceptualization, bulk of the data preparation, software implementation and debugging in R & Python;
  • Alex H. Turnbull: conceptualization, some data preparation, proof-reading and debugging of software.

License

MIT License as separately included.
In short, do what you want with this script, but refer to the original authors when you use or develop this code.

License

MIT License as separately included.
In short, do what you want with this script, but refer to the original authors when you use or develop this code.

FewBit β€” a library for memory efficient training of large neural networks

FewBit FewBit β€” a library for memory efficient training of large neural networks. Its efficiency originates from storage optimizations applied to back

24 Oct 22, 2022
[ECCV'20] Convolutional Occupancy Networks

Convolutional Occupancy Networks Paper | Supplementary | Video | Teaser Video | Project Page | Blog Post This repository contains the implementation o

622 Dec 30, 2022
Rasterize with the least efforts for researchers.

utils3d Rasterize and do image-based 3D transforms with the least efforts for researchers. Based on numpy and OpenGL. It could be helpful when you wan

Ruicheng Wang 8 Dec 15, 2022
πŸš— INGI Dakar 2K21 - Be the first one on the finish line ! πŸš—

πŸš— INGI Dakar 2K21 - Be the first one on the finish line ! πŸš— This year's first semester Club Info challenge will put you at the head of a car racing

ClubINFO INGI (UCLouvain) 6 Dec 10, 2021
An Open-Source Tool for Automatic Disease Diagnosis..

OpenMedicalChatbox An Open-Source Package for Automatic Disease Diagnosis. Overview Due to the lack of open source for existing RL-base automated diag

8 Nov 08, 2022
This repository includes the official project for the paper: TransMix: Attend to Mix for Vision Transformers.

TransMix: Attend to Mix for Vision Transformers This repository includes the official project for the paper: TransMix: Attend to Mix for Vision Transf

Jie-Neng Chen 130 Jan 01, 2023
The codes and related files to reproduce the results for Image Similarity Challenge Track 1.

ISC-Track1-Submission The codes and related files to reproduce the results for Image Similarity Challenge Track 1. Required dependencies To begin with

Wenhao Wang 115 Jan 02, 2023
A self-supervised 3D representation learning framework named viewpoint bottleneck.

Pointly-supervised 3D Scene Parsing with Viewpoint Bottleneck Paper Created by Liyi Luo, Beiwen Tian, Hao Zhao and Guyue Zhou from Institute for AI In

63 Aug 11, 2022
Official Pytorch and JAX implementation of "Efficient-VDVAE: Less is more"

The Official Pytorch and JAX implementation of "Efficient-VDVAE: Less is more" Arxiv preprint Louay Hazami   Β·   Rayhane Mama   Β·   Ragavan Thurairatn

Rayhane Mama 144 Dec 23, 2022
Image Restoration Toolbox (PyTorch). Training and testing codes for DPIR, USRNet, DnCNN, FFDNet, SRMD, DPSR, BSRGAN, SwinIR

Image Restoration Toolbox (PyTorch). Training and testing codes for DPIR, USRNet, DnCNN, FFDNet, SRMD, DPSR, BSRGAN, SwinIR

Kai Zhang 2k Dec 31, 2022
Repository for code and dataset for our EMNLP 2021 paper - β€œSo You Think You’re Funny?”: Rating the Humour Quotient in Standup Comedy.

AI-OpenMic Dataset The dataset is available for download via the follwing link. Repository for code and dataset for our EMNLP 2021 paper - β€œSo You Thi

6 Oct 26, 2022
This repository contains the code used to quantitatively evaluate counterfactual examples in the associated paper.

On Quantitative Evaluations of Counterfactuals Install To install required packages with conda, run the following command: conda env create -f requi

Frederik HvilshΓΈj 1 Jan 16, 2022
🐦 Quickly annotate data from the comfort of your Jupyter notebook

🐦 pigeon - Quickly annotate data on Jupyter Pigeon is a simple widget that lets you quickly annotate a dataset of unlabeled examples from the comfort

Anastasis Germanidis 647 Jan 05, 2023
Context Axial Reverse Attention Network for Small Medical Objects Segmentation

CaraNet: Context Axial Reverse Attention Network for Small Medical Objects Segmentation This repository contains the implementation of a novel attenti

401 Dec 23, 2022
Manifold-Mixup implementation for fastai V2

Manifold Mixup Unofficial implementation of ManifoldMixup (Proceedings of ICML 19) for fast.ai (V2) based on Shivam Saboo's pytorch implementation of

Nestor Demeure 16 Jul 25, 2022
Yet Another Robotics and Reinforcement (YARR) learning framework for PyTorch.

Yet Another Robotics and Reinforcement (YARR) learning framework for PyTorch.

Stephen James 51 Dec 27, 2022
code for paper "Does Unsupervised Architecture Representation Learning Help Neural Architecture Search?"

Does Unsupervised Architecture Representation Learning Help Neural Architecture Search? Code for paper: Does Unsupervised Architecture Representation

39 Dec 17, 2022
Official PyTorch implementation of "VITON-HD: High-Resolution Virtual Try-On via Misalignment-Aware Normalization" (CVPR 2021)

VITON-HD β€” Official PyTorch Implementation VITON-HD: High-Resolution Virtual Try-On via Misalignment-Aware Normalization Seunghwan Choi*1, Sunghyun Pa

Seunghwan Choi 250 Jan 06, 2023
An implementation of based on pytorch and mmcv

FisherPruning-Pytorch An implementation of Group Fisher Pruning for Practical Network Compression based on pytorch and mmcv Main Functions Pruning f

Peng Lu 15 Dec 17, 2022
Social Network Ads Prediction

Social network advertising, also social media targeting, is a group of terms that are used to describe forms of online advertising that focus on social networking services.

Khazar 2 Jan 28, 2022