topic intro model phil
TRANSCRIPT
-
8/8/2019 Topic Intro Model Phil
1/41
1
for t > t 0 ,
C ( x, t ) = C 02
exp(U w) x
2 D
erfc
x wt 2 Dt
erfc
x w t t 0( )2 D t t 0( )
+exp (U +w) x2 D erfc x +wt 2 Dt erfc
x +w t t 0( )2 D t t 0( )
for 0 < t t 0 ,
C ( x, t ) = C 02 exp (U w) x2 D
erfc x wt 2 Dt
+exp (U +w) x2 D
erfc x +wt 2 Dt
where w =U 1+ 4 kDU 2
d 2
2
x Skc x
c D+ xcv-=
t c
-
8/8/2019 Topic Intro Model Phil
2/41
2
1 Introduction
Lesson objectives:To understand conceptually the steps in modeling:
1. Define the problem2. Conceptualization of system--determine what
is known, define system3. Selection or development of mathematical
description (model)--important to understandassumptions, data requirements and model
limitations4. Model calibration--parameter adjustment5. Evaluation of model results--do they make
sense??
-
8/8/2019 Topic Intro Model Phil
3/41
3
Modeling objectives
Increase understanding--which phenomena areimportant, may provide insight into fundamentalbehavior
Predict effects of mitigation, remediation efforts
Estimate exposure => health & environmental risk
Model Identification
-
8/8/2019 Topic Intro Model Phil
4/41
4
Modeling Protocol
-
8/8/2019 Topic Intro Model Phil
5/41
5
Developing conceptual models
1. Establish the PURPOSE of the model.Determine what questions are to be answered.Defines what equations need to be selected and solved.Determines what type of data is necessary.
2. Define the problem : contaminant(s), locationDefine the evaluative environment Source: steady? Plant discharge, leaking tank, old spill? Relevant phases: what media (air, water, soil) are
affected? Properties of media important Model boundaries: scale? Soil grains -> globe
Modeling Protocol
-
8/8/2019 Topic Intro Model Phil
6/41
6
Example : someone has dumped their houseboat waste into Lake
Powell. Health officials are concerned about immediate and futureeffects on human health. Define the evaluative environment.
soilair
water
sediment
What characteristics of environment could be important?
Air water soil/sedimentP, T, wind velocity density grain density(depends on altitude) mixing, stratification organic matterMixing, stratification T (P) permeabilityAerosol-deposition rates surface tension porosity
flowpHionic strengthredox potentialparticulate matter
-
8/8/2019 Topic Intro Model Phil
7/41
7
Develop a CONCEPTUAL MODEL of the system.
Must gather and integrate field data.i.e. Need to determine the type of aquifer, its units, andboundaries.
Often includes the conceptual water budget.It is important to visit the field site during this step!
-
8/8/2019 Topic Intro Model Phil
8/41
8
Select proper GOVERNING EQUATIONS andCOMPUTER CODE
Need equations and code to model your specific site.Should verify equations and code against known analyticalsolutionsor to modeling problems in similar settings.
DESIGN the model
Convert conceptual model to the form necessary for theequations and code.
Includes descritization in time and space.Must set boundary and initial conditions.Must select initial parameters to be used in model.
-
8/8/2019 Topic Intro Model Phil
9/41
9
CALIBRATE the model
Compare model to experimental data (calibration data set) Calibrate model based on calibration data set--model
coefficients, rate constants initially chosen from lab studiesor literature--tune within a range of observed or reportedvalues, using some mathematically defined criteria (e.g.minimizing sum of squared errors)
Performance criteria--how much error is acceptable? Forboth calibration and verification-can be tight or loose; define
before calibration or verification
-
8/8/2019 Topic Intro Model Phil
10/41
10
Verify the ModelOnce you have a model you must measure (collect data) to
verify the model(known as model validation).All Models must be verified and validated.All Models must be verified and validated.
Model VerificationModel VerificationModel VerificationModel Verification
Model verification refers to the process of removing computationalor conceptual errors from a model. This process is an attempt tomake the model internally consistent, and is usually performed bythe developers of the model before it is accessible to the general
public. Even after a model is in use, model users may occasional finderrors or inconsistencies. If this information is communicated to thedevelopers, they can improve later versions of the model.
-
8/8/2019 Topic Intro Model Phil
11/41
11
VALIDATE the model
See if the model can simulate a second set of field data correctly
(validation data set.)Must specify the purpose of model and conditions and range in
which it is valid (all have limitations)
Use the model to PREDICT
Use the calibrated model to determine how it responds to futureevents.Must estimate future conditions.
-
8/8/2019 Topic Intro Model Phil
12/41
12
Sensitivity and uncertainty analysis
Determine how the model responds to uncertainty inparameter values. May need to do for different stress periods. Can be a lot of uncertainty in values for future conditions.
Determine SENSITIVITY to future parameter uncertainty sensitivity: which input parameters make the mostdifference in the model output uncertainty: which input parameters should we knowmore precisely to make the model more precise?
Both are important
outputinchangeparameterinchange
:asdefinedoftenysensitivit
-
8/8/2019 Topic Intro Model Phil
13/41
13
Sensitivity analysis
Definitions vary. In most studies, sensitivity analysis is thestudy of model properties through - not necessarily realisticallysized - changes in the input variables and the analysis of itseffect on model outputs. The questions addressed are for
instance:
whether or not some output is affected at all by someinput
continuity, differentiability, monotonic increase ordecrease of the model's response to input variation
Most of the variation of outputs is generally caused by asmall number of inputs.
-
8/8/2019 Topic Intro Model Phil
14/41
14
UncertaintyIn this context: imperfect knowledge regarding aspects of a model.Uncertainty regarding model variables is usually specified by aprobability distribution or by a sample of measured values (an
empirical probability distribution); sometimes it is specified by a setof possible values. We adhere to the probabilistic concept of uncertainty, and we use variances as measure of uncertainty.
-
8/8/2019 Topic Intro Model Phil
15/41
15
Sources of uncertainty
Uncertainty exists at the level of inputs and output of the model.Uncertainty of model formulation also exists. We will assumethat the model is deterministic, and that uncertainties aresolely introduced via the inputs.
Input uncertainty is caused by natural variation (e.g. weather, soilor water variation) as well as by imperfection of data. Although thecauses of uncertainties may differ, their effect is the same, namelyuncertainty about model outputs. Up to the modeler whether or notto incorporate natural variation in the model; the choice dependsalso on the spatial or temporal scale at which the model is used.
The input uncertainty of different parameters may containcorrelations caused by biological or physical mechanisms, e.g.correlation between photosynthesis rate during the day or night, or
between weather at two consecutive days.
-
8/8/2019 Topic Intro Model Phil
16/41
16
PRESENT RESULTS of model and model design
Probably one of the most important steps.You may have a great model, but you need to show to others.Difficult to do for laypersons.
Remember that modeling is an iterative process.After the above steps are completed, it is important to re-examine the model,Find where it is successful and where it is less than satisfactory,
then improve it.
-
8/8/2019 Topic Intro Model Phil
17/41
17
This is science and philosophy, like thescientific method.
It has changed the way science is done and isthe basis of a modern technologically basedapproach.
-
8/8/2019 Topic Intro Model Phil
18/41
18
Some Thoughts on Modeling the Environment &Models used in Environmental Evaluation
Essential skills for successful model development.
1. Identifying the problem variables accurately.
2. Constructing appropriate relations between these variables.
3. Taking measurements judging the size of quantities that aresignificant and the limitations for these relationships. Set limits.
-
8/8/2019 Topic Intro Model Phil
19/41
19
4. Collecting data, decide how to use them and test yourmodel before your finalize it. Modify the model with dataand with additional theoretical relationships that you findthrough this process.
5. Know and document the limits of the model by estimating
the parameters within the model that cannot be measured orcalculated from data.
6. Limit the use of the model to its functional and appropriatecapabilities excluding the inappropriate use of the model.
-
8/8/2019 Topic Intro Model Phil
20/41
20
It has been suggested that canisters of nuclear waste can safely be driven into theocean floor and stored without contaminating large regions of ocean.Model Proposed by Researcher
Assumptions made by the model
1. The canisters will penetrate the ocean floor to a minimum depth of 50 Meters.2. No disturbance of the ocean water will be caused by the presence of thecanisters.3. Canisters will last for over 1,000 years.4. After 1,000 years, leaching will occur.5. A diffusion model was constructed to distribute the radioactive leachedisotopes. Taking into consideration exclusively brownian movement anddiffusion aspects of the dispersion in the water.
Example:Ocean Disposal of Nuclear Waste
-
8/8/2019 Topic Intro Model Phil
21/41
21
Q uestions:
How was the model constructed?What are the important variables and parameters?What are the relationships?
What is the mathematical relationship? (an example of atypical diffusion algorithm)
-
8/8/2019 Topic Intro Model Phil
22/41
22
The process of solute movement are commonly described by the advection-
dispersion equation.
c/ t = D m(2c/ x2) - v c/ x
Where: c/ t is the change in concentration with the change in time, andc is solute concentration
t is time
D is the dispersion coefficient ( D = Do+ ev )where: e is the coefficient of solute dispersivity in soil, and Do is themolecular diffusion coefficient
v is pore water velocity (given by water flux (q) divided by the volumetricwater constant (theta)
Ref. Modeling Chemical Transport in Soils: Natural and AppliedContaminants, Hossein Chadiri and Calvin Rose, Lewis Publishers, pg. 147,1992.
-
8/8/2019 Topic Intro Model Phil
23/41
23
Then, the following steps were taken and questions asked:
Model Verification, Evaluation, Validation, and Limitation:
Were the assumptions valid?Is this an appropriate model?If it was not a good model, why and how would you improve it?
-
8/8/2019 Topic Intro Model Phil
24/41
24
Model Type
Deterministic Stochastic
Steady State non-Steady State Steady State non-Steady State
Conservative non-Conservative Conservative non-Conservative
-
8/8/2019 Topic Intro Model Phil
25/41
25
Some model types
Deterministic : a partial differential is solved, numerically oranalytically, for a given set of input values, atmosphere,stream, lake or aquifer parameters, and boundary conditions.
The resulting output variable has a specific value at a givenplace in the system. There is a fixed relationship between inputand output.
Steady state : system does not vary with time. Inputs are constantand system eventually reaches some equilibrium condition.a. conservative parameter: material doesn't react or decay.b. nonconservative: waste undergoes reactions, transformations,decay.
Dynamic : system changes with time due to changing conditionsor inputs.a. conservativeb. nonconservative
-
8/8/2019 Topic Intro Model Phil
26/41
26
Stochastic : a model for which there is a statistical
uncertainty in the value of the output variables due touncertainties in the system parameters, input parameters,or errors in measurement. Models allow for randomfluctuations. Uses mean and standard deviations of parameters.
Steady state , conservative and nonconservative
Dynamic , conservative and nonconservative
-
8/8/2019 Topic Intro Model Phil
27/41
27
Model Type
Deterministic Stochastic
Steady State non-Steady State Steady State non-Steady State
Conservative non-Conservative Conservative non-Conservative
-
8/8/2019 Topic Intro Model Phil
28/41
28
Contaminant Transport Mass Balance Advective transport Diffusion and
dispersion Volatilization Adsorption
Biodegradationprocesses Chemical Reaction
-
8/8/2019 Topic Intro Model Phil
29/41
29
1.2.4 Define fate and transport phenomena
physical basic mass balance, reactor theoryadvection : transport of contaminants due to bulk flowof water or air diffusion : spread of contaminants due to randommolecular motion dispersion : spread of contaminants due to mechanicalmixing
Can have transport between phases (aqueous, vapor,solid), transport within a phase, complete mixing withina phase (Topics 4 & 5)
chemical and biological rxns (Topic 2)
-
8/8/2019 Topic Intro Model Phil
30/41
30
Subdivisions of Transport Process
Point Sourcesintentional
well defined sourceboundary
Nonpoint sourcesFugitive emissions
unintentional emissionsIll-defined source boundary
Loading Advective
physical movementof contaminants
Dispersive
spreading to down gradientmixing due to turbulence etc.
Diffusivemovement from higherto lower concentration
Transport
Reactions
HomogeneousSingle phase
HeterogeneousMultiple phases
Mass Balance
-
8/8/2019 Topic Intro Model Phil
31/41
31
Sources of Contaminate Loading
Industrial spills,discharges and leaks
Surface impoundments Storage tanks and
pipes Landfills Burial areas and
dumps
Injection wells
-
8/8/2019 Topic Intro Model Phil
32/41
32
Areas of Contamination Surface soils
Subsurface soils Shallow ground water Deep ground water Vapors above water
table Drinking water wells Receiving
streams/lakes
Air
-
8/8/2019 Topic Intro Model Phil
33/41
33
Mechanisms of Contamination
ConfiningUnit
Water table
SalineWater
Lateralintrusion ofsaline water
Ocean
Municipalwater well
Abandonedoil well
DeepAquifer
pond
Infiltration of
pesticides andfertilizers fromfarmlands
Brine leakage fromruptured well casing
septic tankleakageFreshwater
Accidentalfuel spill
Municipallandfill
Leakage from
hazardouswaste siteContaminated
shallowwell
Leakingpetroleum
tank
ConfiningUnit
-
8/8/2019 Topic Intro Model Phil
34/41
34
Leachate togroundwater
Landfill
Volatilization
To upper atmosphere
Wind dispersed contaminants
Precipitation
Inflow
Sewer
PhotochemicalsSunlight
Volatilization
PhotolysisHydrolysisAcid-base equilibria
Sediment
Adsorption/ desorption
Sedimentation
Bioaccumulation
Biodegradation
Leachate togroundwater
Sun
-
8/8/2019 Topic Intro Model Phil
35/41
35
1.2.5 Define the state of the system
Steady or unsteady stateSteady state : no change of mass &/or concentration at
a point with timedmdt
=0, dC dt
=0
m or C
t (at a point )
Opposite is unsteady state,
or Dynamic
-
8/8/2019 Topic Intro Model Phil
36/41
36
Equilibrium or nonequilibrium (partitioning between 2 phases)
Chemical equilibrium : net rate of mass transfer betweenphases is zero (or rate of forward and backward reaction is 0)
Ratio of concentrations in phases = a specific constant calledthe equilibrium partition coefficient = K ij
AirCA
WaterCw
C AC w
=K AW at equilibriumC
t
CA
Cw
Concs. change,ratio is constant
-
8/8/2019 Topic Intro Model Phil
37/41
37
Example :Given the system is at equilibriumand steady state;
each phase is well mixed.
C AC w
=20
AirV=100 m 3
CA=8 mol O 2 /m 3
WaterV=100 m 3
Cw=0.4 mol O 2 /m 3
a. AirCA=8 mol
O2 /m 3
WaterCw=0.4 mol
O2 /m 3
CA,in = 8 molO2 /m 3
Cw,in = 0.4 molO2 /m 3
CA,out =8 molO2 /m 3
Cw,out =0.4 molO2 /m 3
All concs. constantEquilibrium? Steady State?
b. Air
Water
CA,in = 8 mol
O2 /m 3Cw,in = 0.1 mol
O2 /m 3
(both constant)
CA,out =6 molO2 /m 3
Cw,out =0.2 molO2 /m 3
(constant)
Equilibrium? no--ratio 20Steady State?
-
8/8/2019 Topic Intro Model Phil
38/41
38
c.
Both C ins vary
Both C outsvary butratio isalways 20
Example (cont.)
d.
Equilibrium? (notethat influent streams do NOT
need to be at equilibrium forthe system to be at equilibrium)Steady State? no
Equilibrium? noSteady State? no
Both C ins vary
Both C outsvary andratio 20
-
8/8/2019 Topic Intro Model Phil
39/41
39
1.3 Developing mathematical models
Mass balance approach (engineering fundamental)--mass cannotbe created or destroyed (except nuclear rxns), but it can betransferred or transformed
Define control volume (evaluative environment) Define inputs and outputs across and within boundaries =>mass balance equation (mathematical model)
Outputs
Boundary
Inputs Accumulation
Decay
Accumulation or rate of change of storage = input - output +reaction (+ generation - consumption)
-
8/8/2019 Topic Intro Model Phil
40/41
40
Example (will cover in more depth in Topic 4): closed, steady
state system--no mass flow into or out of system
AirVA, C A
WaterVw, C w
SedimentVs, C s
Total mass in this evaluative environment = sumof mass in each phase
m total = C AVA + C wVw + C sVs
Given C w, V is-- relate C A and C s to C w by
equilibrium partition coefficients--Henrys law K aw = C A /C w--sediment/water partition coeff sw = C s /C w
m total = K awCwVA + C wVw + swCwVs= (K awVA + V w + swV s) C w
-
8/8/2019 Topic Intro Model Phil
41/41
41
Well look at 2 good models with sound physical basis
completely mixed compartment (box) modelNot spatially dependent--concentration is uniform in
each phase
advection/dispersion modelsChanges in concentration with both time and space