Making A Single Prediction¶

In [1]:

from ml4pd import components
from ml4pd.streams import MaterialStream
from ml4pd.aspen_units import Distillation

from ml4pd import components from ml4pd.streams import MaterialStream from ml4pd.aspen_units import Distillation

Specify Molecules in System¶

Before any predictions can be done, a list of participating molecules (identified via their iupac names) must be added to components, similar to how Aspen operates. components will then gather features for these molecules and streams will check input molecules against this list to make sure they've been declared. Under the hood, components uses the packages thermo & rdkit to gather additional identifiers like iupac names, smiles strings, and CAS # and also molecular features.

In [2]:

components.set_components(['acetone', '2-butanone'])

components.set_components(['acetone', '2-butanone'])

Sometimes thermo might mis-identify the input molecules, so it's good practice to check by printing components like so.

In [3]:

components

components

Out[3]:

         name         iupac    smiles      cas
0     acetone  propan-2-one   CC(=O)C  67-64-1
1  2-butanone   butan-2-one  CCC(=O)C  78-93-3

Create Streams¶

Streams can be lazily initiliazed without any parameters. But they need to be called to put togther data for ML. A few notes on how to specify data for streams:

• Information on molecules & flowrates can only specified during call, not initialization.
• Molecules & flowrates have to be in the form of dictionaries (see code).
• All other parameters can be specified during initiliazaiton & call.
• If the same parameter is specified during initilization and call, it will be overwritten with call.

In [4]:

molecules = {"name_A": ["acetone"], "name_B": ["2-butanone"]}
flowrates = {"flowrate_A": [0.5], "flowrate_B": [0.5]}

feed_stream_1 = MaterialStream(vapor_fraction=0, pressure=2)(molecules=molecules, flowrates=flowrates)
feed_stream_2 = MaterialStream()(molecules=molecules, flowrates=flowrates, vapor_fraction=0, pressure=2)

molecules = {"name_A": ["acetone"], "name_B": ["2-butanone"]} flowrates = {"flowrate_A": [0.5], "flowrate_B": [0.5]} feed_stream_1 = MaterialStream(vapor_fraction=0, pressure=2)(molecules=molecules, flowrates=flowrates) feed_stream_2 = MaterialStream()(molecules=molecules, flowrates=flowrates, vapor_fraction=0, pressure=2)

Data used for ML held by a stream is stored in the data attribute.

In [5]:

feed_stream_1.data

feed_stream_1.data

Out[5]:

	name_A	name_B	smiles_A	iupac_A	cas_A	MaxEStateIndex_A	MinEStateIndex_A	qed_A	MolWt_A	HeavyAtomMolWt_A	NumValenceElectrons_A	MaxPartialCharge_A	MinPartialCharge_A	FpDensityMorgan1_A	FpDensityMorgan2_A	FpDensityMorgan3_A	BCUT2D_MWHI_A	BCUT2D_MWLOW_A	BCUT2D_CHGHI_A	BCUT2D_CHGLO_A	BCUT2D_LOGPHI_A	BCUT2D_LOGPLOW_A	BCUT2D_MRHI_A	BCUT2D_MRLOW_A	BalabanJ_A	BertzCT_A	Chi0_A	Chi0n_A	Chi0v_A	Chi1_A	Chi1n_A	Chi1v_A	Chi2n_A	Chi2v_A	Chi3n_A	Chi3v_A	Chi4n_A	Chi4v_A	HallKierAlpha_A	Ipc_A	Kappa1_A	Kappa2_A	Kappa3_A	LabuteASA_A	PEOE_VSA1_A	PEOE_VSA10_A	PEOE_VSA14_A	PEOE_VSA2_A	PEOE_VSA4_A	PEOE_VSA6_A	PEOE_VSA7_A	PEOE_VSA8_A	SMR_VSA1_A	SMR_VSA10_A	SMR_VSA5_A	SlogP_VSA2_A	SlogP_VSA3_A	SlogP_VSA5_A	TPSA_A	EState_VSA10_A	EState_VSA2_A	EState_VSA3_A	EState_VSA4_A	EState_VSA5_A	EState_VSA6_A	EState_VSA7_A	EState_VSA8_A	EState_VSA9_A	VSA_EState2_A	VSA_EState5_A	VSA_EState7_A	VSA_EState8_A	FractionCSP3_A	HeavyAtomCount_A	NHOHCount_A	NOCount_A	NumHAcceptors_A	NumHeteroatoms_A	NumRotatableBonds_A	MolLogP_A	MolMR_A	fr_C_O_A	fr_C_O_noCOO_A	fr_ketone_A	fr_ketone_Topliss_A	fr_unbrch_alkane_A	tm_A	tb_A	tc_A	pc_A	vc_A	z_A	rhoc_A	acenttric_factor_A	triple_temp_A	triple_pres_A	heat_vaporization_A	heat_fusion_A	stockmayer_param_A	solubility_param_A	parachor_A	smiles_B	iupac_B	cas_B	MaxEStateIndex_B	MinEStateIndex_B	qed_B	MolWt_B	HeavyAtomMolWt_B	NumValenceElectrons_B	MaxPartialCharge_B	MinPartialCharge_B	FpDensityMorgan1_B	FpDensityMorgan2_B	FpDensityMorgan3_B	BCUT2D_MWHI_B	BCUT2D_MWLOW_B	BCUT2D_CHGHI_B	BCUT2D_CHGLO_B	BCUT2D_LOGPHI_B	BCUT2D_LOGPLOW_B	BCUT2D_MRHI_B	BCUT2D_MRLOW_B	BalabanJ_B	BertzCT_B	Chi0_B	Chi0n_B	Chi0v_B	Chi1_B	Chi1n_B	Chi1v_B	Chi2n_B	Chi2v_B	Chi3n_B	Chi3v_B	Chi4n_B	Chi4v_B	HallKierAlpha_B	Ipc_B	Kappa1_B	Kappa2_B	Kappa3_B	LabuteASA_B	PEOE_VSA1_B	PEOE_VSA10_B	PEOE_VSA14_B	PEOE_VSA2_B	PEOE_VSA4_B	PEOE_VSA6_B	PEOE_VSA7_B	PEOE_VSA8_B	SMR_VSA1_B	SMR_VSA10_B	SMR_VSA5_B	SlogP_VSA2_B	SlogP_VSA3_B	SlogP_VSA5_B	TPSA_B	EState_VSA10_B	EState_VSA2_B	EState_VSA3_B	EState_VSA4_B	EState_VSA5_B	EState_VSA6_B	EState_VSA7_B	EState_VSA8_B	EState_VSA9_B	VSA_EState2_B	VSA_EState5_B	VSA_EState7_B	VSA_EState8_B	FractionCSP3_B	HeavyAtomCount_B	NHOHCount_B	NOCount_B	NumHAcceptors_B	NumHeteroatoms_B	NumRotatableBonds_B	MolLogP_B	MolMR_B	fr_C_O_B	fr_C_O_noCOO_B	fr_ketone_B	fr_ketone_Topliss_B	fr_unbrch_alkane_B	tm_B	tb_B	tc_B	pc_B	vc_B	z_B	rhoc_B	acenttric_factor_B	triple_temp_B	triple_pres_B	heat_vaporization_B	heat_fusion_B	stockmayer_param_B	solubility_param_B	parachor_B	feed0_flowrate_A	feed0_flowrate_B	feed0_temperature	feed0_vapor_fraction	feed0_pressure	feed0_comp_no
0	acetone	2-butanone	CC(=O)C	propan-2-one	67-64-1	9.444444	0.166667	0.398237	58.08	52.032	24	0.126268	-0.300344	1.5	1.5	1.5	16.136528	10.550822	1.619125	-1.557836	1.500722	-1.691321	5.717069	-0.114493	2.803039	26.264663	3.57735	2.908248	2.908248	1.732051	1.204124	1.204124	0.908248	0.908248	0.0	0.0	0.0	0.0	-0.33	3.245112	3.67	1.044532	6.883958	25.630657	4.794537	5.783245	0.0	0.0	0.0	0.0	13.847474	0.0	4.794537	5.783245	13.847474	5.783245	4.794537	13.847474	17.07	4.794537	5.783245	0.0	0.0	13.847474	0.0	0.0	0.0	0.0	9.444444	0.166667	0.0	3.055556	0.666667	4	0	1	1	1	0	0.5953	16.355	1	1	1	1	0	178.35	329.23	508.1	4700000.0	0.000213	0.23697	272.672019	0.309	176.732207	3.047365	29122.034793	5770.0	332.97	19638.941183	0.000029	CCC(=O)C	butan-2-one	78-93-3	9.8125	0.25463	0.451051	72.107	64.043	30	0.129065	-0.300042	1.8	2.0	2.0	16.137114	10.3607	1.764362	-1.711045	1.710825	-1.797012	5.744225	-0.116199	2.847379	38.912609	4.284457	3.615355	3.615355	2.270056	1.764784	1.764784	1.055568	1.055568	0.497891	0.497891	0.0	0.0	-0.33	9.651484	4.67	1.94248	3.67	31.995599	4.794537	5.783245	0.0	0.0	0.0	6.923737	6.923737	6.420822	4.794537	5.783245	20.268296	5.783245	4.794537	20.268296	17.07	4.794537	5.783245	6.420822	0.0	0.0	6.923737	6.923737	0.0	0.0	9.8125	0.25463	0.666667	3.43287	0.75	5	0	1	1	1	1	0.9854	20.972	1	1	1	1	0	186.35	352.75	536.7	4207000.0	0.000267	0.25172	270.058876	0.329	186.46881	1.735133	31555.836658	8390.0	415.48	18879.750645	0.000036	0.5	0.5	None	0.0	2.0	2

Create Columns¶

Like streams, columns can be lazily initialized but has to be called to produce predictions. A few notes on how to specify data for columns:

• streams can only be fed into columns during call.
• All other parameters can be specified during initilization or call.
• If the same parameter is specified during initilization and call, it will be overwritten with call.

In [6]:

dist_col_1 = Distillation(reflux_ratio=0.1, boilup_ratio=0.1, pressure=2, no_stages=4, feed_stage=2)
dist_col_2 = Distillation()

bott_1, dist_1 = dist_col_1(feed_stream_1)
bott_2, dist_2 = dist_col_2(reflux_ratio=0.1, boilup_ratio=0.1, pressure=2, no_stages=4, feed_stage=2, feed_stream=feed_stream_2)

dist_col_1 = Distillation(reflux_ratio=0.1, boilup_ratio=0.1, pressure=2, no_stages=4, feed_stage=2) dist_col_2 = Distillation() bott_1, dist_1 = dist_col_1(feed_stream_1) bott_2, dist_2 = dist_col_2(reflux_ratio=0.1, boilup_ratio=0.1, pressure=2, no_stages=4, feed_stage=2, feed_stream=feed_stream_2)

Inspect Results¶

Columns will return stream objects filled with ML results and also fill itself with ML results. The types & number of stream objects, as well as the results available via ML depend on the type of column & type of data specified.

In [7]:

bott_1.flow

bott_1.flow

Out[7]:

	flowrate_A	flowrate_B
0	0.443523	0.476227

In [8]:

bott_2.temperature

bott_2.temperature

Out[8]:

[88.45621634567827]

In [9]:

dist_col_1.condensor_duty

dist_col_1.condensor_duty

Out[9]:

[-1682.6839428518751]