gnome.weatherers.cleanup

oil removal from various cleanup options add these as weatherers

Module Contents

Classes

RemoveMass	create a mixin for mass removal. These methods are used by CleanUpBase and
CleanUpBase	Just need to add a few internal methods for Skimmer + Burn common code
Skimmer	Just need to add a few internal methods for Skimmer + Burn common code
Burn	Just need to add a few internal methods for Skimmer + Burn common code
ChemicalDispersion	Just need to add a few internal methods for Skimmer + Burn common code

class gnome.weatherers.cleanup.RemoveMass

Bases: object

create a mixin for mass removal. These methods are used by CleanUpBase and also by manual_beaching.

valid_vol_units

valid_mass_units

_get_mass(substance, amount, units)

return ‘amount’ in units of ‘kg’ for specified substance uses the density corresponding with API temperature

_set__timestep(time_step, model_time)

For cleanup operations we may know the start time pretty precisely. Use this to set _timestep to less than time_step resolution. Mostly done for testing right now so if XXX amount is skimmed between active start and active stop, the rate * duration gives the correct amount. Object must be active before invoking this, else self._timestep will give invalid results

prepare_for_model_step(sc, time_step, model_time)

Do sub timestep resolution here so numbers add up correctly Mark LEs to be skimmed - do them in order right now. Assume all LEs that are released together will be skimmed together since they would be closer to each other in position.

Assumes: there is more mass in water than amount of mass to be skimmed. The LEs marked for Skimming are marked only once - code checks to see if any LEs are marked for skimming and if none are found, it marks them.

class gnome.weatherers.cleanup.CleanUpBase(efficiency=1.0, **kwargs)

Bases: RemoveMass, gnome.weatherers.Weatherer

Just need to add a few internal methods for Skimmer + Burn common code Currently defined as a base class.

add ‘frac_water’ to array_types and pass **kwargs to base class __init__ using super

property efficiency

• Efficiency can be None since it indicates that we use wind to compute efficiency.

• If efficiency is not None, it must be a number greater than or equal to 0.0 and less than or equal to 1.0.

_update_LE_status_codes(sc, new_status, substance, mass_to_remove, oilwater_mix=True)

Need to mark LEs to ‘new_status’. It updates the ‘fate_status’ for ‘surface_weather’ LEs. Mark LEs based on mass. Mass to remove is assumed to be the oil/water mixture by default (oilwater_mix=True) so we need to find the oil_amount given the water_frac:

volume = sc[‘mass’]/API_density

(1 - sc[‘frac_water’]) * oil_water_vol = volume

oil_water_vol = volume / (1 - sc[‘frac_water’])

Now, do a cumsum of oil_water_mass and find where

np.cumsum(oil_water_vol) >= vol_to_remove

and change the status_codes of these LEs. Can just as easily multiple everything by API_density to get

np.cumsum(oil_water_mass) >= mass_to_remove

mass_to_remove = sc[‘mass’] / (1 - sc[‘frac_water’])

This is why the input is ‘mass_to_remove’ instead of ‘vol_to_remove’ - less computation

Note:

For ChemicalDispersion, the mass_to_remove is not the mass of the oil/water mixture, but the mass of the oil. Use the oilwater_mix flag to indicate this is the case.

_avg_frac_oil(data)

find weighted average of frac_water array, return (1 - avg_frac_water) since we want the average fraction of oil in this data

class gnome.weatherers.cleanup.Skimmer(amount=0, units=None, water=None, **kwargs)

Bases: CleanUpBase

Just need to add a few internal methods for Skimmer + Burn common code Currently defined as a base class.

initialize Skimmer object - calls base class __init__ using super() active_range is required cleanup operations must have a valid datetime - cannot use -inf and inf active_range is used to get the mass removal rate

property units

return units for amount skimmed

_schema

_ref_as = 'skimmer'

_req_refs = ['water']

_validunits(value)

checks if units are either valid_vol_units or valid_mass_units

prepare_for_model_run(sc)

no need to call base class since no new array_types were added

prepare_for_model_step(sc, time_step, model_time)

Do sub timestep resolution here so numbers add up correctly Mark LEs to be skimmed - do them in order right now. Assume all LEs that are released together will be skimmed together since they would be closer to each other in position.

Assumes: there is more mass in water than amount of mass to be skimmed. The LEs marked for Skimming are marked only once - code checks to see if any LEs are marked for skimming and if none are found, it marks them.

_mass_to_remove(substance)

use density at 15C, ie corresponding with API to do mass/volume conversion

weather_elements(sc, time_step, model_time)

Assumes there is only ever 1 substance being modeled! remove mass equally from LEs marked to be skimmed

class gnome.weatherers.cleanup.Burn(area=None, thickness=None, active_range=(InfDateTime('-inf'), InfDateTime('inf')), area_units='m^2', thickness_units='m', efficiency=1.0, wind=None, water=None, **kwargs)

Bases: CleanUpBase

Just need to add a few internal methods for Skimmer + Burn common code Currently defined as a base class.

Set the area of boomed oil to be burned. Cleanup operations must have a valid datetime for active start, cannot use -inf. Cannot set active stop - burn automatically stops when oil/water thickness reaches 2mm.

Parameters:

• area (float) – area of boomed oil/water mixture to burn

• thickness (float) – thickness of boomed oil/water mixture

• active_range (datetime) – time when the burn starts is the only thing we track. However we give a range to be consistent with all other weatherers.

• area_units (str) – default is ‘m^2’

• thickness_units (str) – default is ‘m’

• efficiency (float) – burn efficiency, must be greater than 0 and less than or equal to 1.0

• wind – gnome.environment.Wind object. Only used to set efficiency if efficiency is None. Efficiency is defined as: 1 - 0.07 * wind.get_value(model_time) where wind.get_value(model_time) is value of wind at model_time

Kwargs passed onto base class:

Parameters:

• name (str) – name of object

• on (bool) – whether object is on or not for the run

property area_units

property active_range

property thickness

property thickness_units

_schema

_ref_as = 'burn'

_req_refs = ['water', 'wind']

valid_area_units

valid_length_units

_log_thickness_warning()

when thickness or thickness_units are updated, check to see that the value in SI units is > _min_thickness. If it is not, then log a warning

prepare_for_model_run(sc)

resets internal _oilwater_thickness variable to initial thickness specified by user and active stop to ‘inf’ again. initializes sc.mass_balance[‘burned’] = 0.0

prepare_for_model_step(sc, time_step, model_time)

1. set ‘active’ flag based on active start, and model_time

2. Mark LEs to be burned - do them in order right now. Assume all LEs that are released together will be burned together since they would be closer to each other in position. Assumes: there is more mass in water than amount of mass to be burned. The LEs marked for Burning are marked only once - during the very first step that the object becomes active

_init_rate_duration(avg_frac_oil=1)

burn duration based on avg_frac_oil content for LEs marked for burn __init__ invokes this to initialize all parameters assuming frac_water = 0.0

_set_burn_params(sc, substance)

Once LEs are marked for burn, the frac_water does not change set burn rate for oil/water thickness, as well as volume burn rate for oil:

If data contains LEs marked for burning, then:

avg_frac_oil = mass_weighed_avg(1 - data[‘frac_water’]) _oilwater_thick_burnrate = 0.000058 * avg_frac_oil _oil_vol_burnrate = _oilwater_thick_burnrate * avg_frac_oil * area

The burn duration is also known if efficiency is constant. However, if efficiency is based on variable wind, then duration cannot be computed.

_set_efficiency(points, model_time)

return burn efficiency either from efficiency attribute or computed from wind

weather_elements(sc, time_step, model_time)

1. figure out the mass to remove for current timestep based on rate and efficiency. Find fraction of total mass and remove equally from all ‘mass_components’ of LEs marked for burning.

2. update ‘mass’ array and the amount burned in mass_balance dict

3. append to _burn_duration for each timestep

class gnome.weatherers.cleanup.ChemicalDispersion(fraction_sprayed, active_range=(InfDateTime('-inf'), InfDateTime('inf')), waves=None, efficiency=1.0, **kwargs)

Bases: CleanUpBase

Just need to add a few internal methods for Skimmer + Burn common code Currently defined as a base class.

another mass removal mechanism. The volume specified gets dispersed with efficiency based on wave conditions.

Parameters:

• volume (float) – volume of oil (not oil/water?) applied with surfactant

• units (str) – volume units

• active_range (2-tuple of datetimes) – Range of datetimes for when the mover should be active

• waves (an object with same interface as gnome.environment.Waves) – waves object - query to get height. It must contain get_value() method. Default is None to support object creation by WebClient before a waves object is defined

Optional Argument: Either efficiency or waves must be set before running the model. If efficiency is not set, then use wave height to estimate an efficiency

Parameters:

efficiency (float between 0 and 1) – efficiency of operation.

remaining kwargs include ‘on’ and ‘name’ and these are passed to base class via super

_schema

_ref_as = 'chem_dispersion'

_req_refs = ['waves']

prepare_for_model_run(sc)

reset _rate to None. It gets set when LEs are marked to be dispersed.

prepare_for_model_step(sc, time_step, model_time)

1. invoke base class method (using super) to set active flag

2. mark LEs for removal

3. set internal _rate attribute for mass removal [kg/sec]

_set_efficiency(points, model_time)

weather_elements(sc, time_step, model_time)

for now just take away 0.1% at every step