Modeling and preparation of activated carbon for methane storage II. Neural network modeling and experimental studies of the activated carbon preparation

Predictable and targeted activation of biomass to carbons with high surface area density and enhanced methane storage capacity

“…This yield is much higher than the yields (i.e., 4 to 20%) obtained when carbonaceous matter that is generated via conventional pyrolysis and hydrothermal carbonisation (HTC) routes is used as the starting material. [6][7][8][9][10][11][12][13][14][15][16][17] We ascribe the higher yield to the fact that flash carbonisation, such as demonstrated here for date seeds and ACDS carbon, generates carbonaceous matter that is relatively resistant to activation with KOH due to having a low O/C ratio. [41][42][43][44] However, the critical mass balance parameter is the overall biomass (date seeds) to activated carbon yield, which is between 25 and 40%.…”

“…To circumvent the challenges of CNG and LNG, methane may be stored as adsorbed natural gas (ANG) via adsorption, at relatively low pressure, onto a suitable adsorbent material such as porous carbon. [6][7][8][9][10][11][12][13][14][15][16][17] Recent research has developed new ways of preparing porous carbons with properties that are useful for good performance in energy-related applications but much of the work still relies on trial and error approaches. [18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] Careful activation of carbonaceous matter can dramatically improve properties for specific applications but there is need for a more predictable synthesis approach.…”

Comparative Study between Regression and Soft Computing Models to Maximize the Methane Storage Capacity of Anthracite-Based Adsorbents

“…Afterward, the trained networks were ready to produce acceptable predictions for unseen (test or validation) data, different from the observed (training) data. The training of the network between the inputs (AC preparation parameters) and hidden layer occurs in an unsupervised fashion, while the training of the network between the hidden and output layers (AC specification) takes place in a supervised style based on the target outputs. − Finally, such a model is said to generalize (interpolate) well within the data range. In other words, generalization performance is computing outputs for some exemplars outside the training set but inside the training domain…”

Modulating the porosity of activated carbons via pre-mixed precursors for simultaneously enhanced gravimetric and volumetric methane uptake

“…Storage of adsorbed methane, typically involving physisorption onto the surface of porous adsorbents, can operate at low pressure conditions and at ambient temperature, both of which can reduce costs and offer improved ease of usage. [4][5][6][7][8][9][10]14,15 Porous carbons are amongst the materials currently being explored for the storage of methane. [4][5][6][7][8][9][10] Porous carbons, and specically activated carbons, have a number of attractive characteristics including controllable porosity, low cost, chemical and mechanical stability, and ease of preparation and handling.…”