MGO Tutorial - Dewatering Scenario
Contents
1. Modf low MT3DMS Call No 6625 Once the optimization simulation is complete the DOS window will close automatically and an Optimization Results window will open Interpreting the results is explained in the following section Interpreting and Utilizing Well Optimization Results Once the simulation is complete the Optimization Results window as shown in the following figure will appear Optimization Results 3 x Decision Variables Objective Function Dewater mgo Name Type Rate m 3 d gt PMP1 Extraction Well 0 PMP2 Extraction Well 0 PMP3 Extraction Well 0 PMP4 Extraction Well 0 PMP5 Extraction Well 0 PMPG Extraction Well 2903 571059038804 PMP7 Extraction Well 4838 615301 152554 PMP8 Extraction Well 0 PMPI Extraction Well 0 PMP10 Extraction Well 4515 10323159875 PMP11 Extraction Well 3225 07373685625 PMP12 Extraction Well 0 PMP13 Extraction Well 0 _ PMP14 Extraction Well 0 E PMP15 Extraction Well 0 F PMP16 Extraction Well 0 INJ1 Injection Well 2903 571059038804 INJ2 Injection Well 4838 615301152554 INJ3 Injection Well 4515 10323159875 INJ4 Injection Well 3225 07373685625 Do you want to import Decision Variables into Visual MODFLOW Running Simulation and Viewing Output 19 The GA solver uses Heuristic algorithms in an attempt to find the optimal solution Consequently your results may vary but they should be similar In the Decision Variables tab we can see that MGO calculated an op2. MGO supports three different solvers for the Optimization process e Genetic Algorithm GA e Simulated Annealing SA e Tabu Search TS Genetic Algorithm for the Solver For this simulation the default Solver settings are fine Solution Settings The default solution settings are sufficient for this simulation Output Settings The default output settings are sufficient for this simulation Run to begin the simulation 18 Defining MGO Input Running Simulation and Viewing Output When you click Run this will launch the MGO engine automatically translate the Visual MODFLOW and MGO input files and launch the necessary numeric engines The MGO engine will run in a DOS window as shown in the following figure Shortcut to Mod Modf low MT3DMS 1 Call No 6168 Average Function Value of Generation 0 56446E 07 Optimal Function Value 0 10000E 07 HHHHHHHHHHHHHHH GA Generation OAAS HHHHHHHHHHHHHHHH Modf low MT3DMS Call No s1 Average Function Value of Generation 0 69525E 07 Optimal Function Value 0 10000E 07 HHH GA Generation G609 Hittitittinniiniiit Modf low MT3DMS1 Call No 6160 Average Function Value of Generation 0 44603E 07 Optimal Function Value 0 10000E 07 HHHHHHHHHHHHHHH GA Generation 6610 HHHHHHHHHHHHHHHH Modf low MT3DMS Call No 81908 Average Function Value of Generation 8 51367E 07 Optimal Function Value 8 10000E 07 HHHHHHHERHHHHHH GA Generation BALA iHiHHHHHHHHHHHHHEH
3. Value Units Row AM Therefore we can conclude that the pumping rates calculated by MGO have satisfied our head constraints We can then consider these well locations and pumping rates as a potential dewatering scenario for this site References Utah State University and Peralta and Associates Inc 2003 SOMOS Simulation Optimization Modeling System Optimization Software for Managing Groundwater Flow Solute Transport and Conjunctive Use User s Manual for Lite and Standard Versions Zheng C and Wang P 2003 MGO Modular Groundwater Optimizer Incorporating MODFLOW MT3DMS Documentation and User s Guide The University of Alabama in cooperation with Groundwater Systems Research Ltd Viewing Visual MODFLOW Output 23 24 References
4. optimization algorithms and capabilities of MGO are provided in the MGO Documentation and User s Guide Zheng and Wang 2003 An electronic copy of this document is included on the Visual MODFLOW installation CD ROM in the Manual folder This tutorial describes how to set up and run a Pumping Well Optimization simulation using Visual MODFLOW Although some details of the required inputs and parameters will be provided in this chapter you are encouraged to read the MGO Documentation and User s Guide in order to better understand the benefits and limitations of Pumping Well Optimization using MGO NOTE Some features described in this tutorial are only available in a Pro or Premium version Well Optimization Limitations In Visual MODFLOW pumping well optimization may only be used under the following conditions e MODFLOW 96 flow engine with PCG2 SIP or SOR solver e MT3DMS transport engine Steady State Flow Background This tutorial illustrates a common dewatering scenario in an excavation project The goal is to lower the water level at a construction site so that a building foundation can be poured The goal of the optimization is to minimize the total amount of water pumped and minimizing the costs while achieving the head constraints Terms and Notations For the purposes of this tutorial the following terms and notations will be used Type type in the given word or value Select click the left mouse button whe
5. 25 1 5000 v Extraction Well PMP2 100 0 5000 32 0 25 1 5000 v Extraction Well PMP3 300 0 5000 32 0 25 1 5000 v Extraction Well PMP4 500 0 5000 32 0 25 1 5000 v Extraction Well PMP5S 100 0 5000 32 0 25 1 5000 v Extraction Well PMP6 1500 0 5000 32 0 25 1 5000 v Extraction Well PMP 1500 0 5000 32 0 25 1 5000 Vv Extraction Well PMP8 500 0 5000 32 0 25 J 5000 v Extraction Well PMPS 100 0 5000 32 0 25 1 5000 v Extraction Well PMP10 1500 0 5000 32 0 25 1 5000 v Extraction Well PMP11 1500 0 5000 32 0 25 1 5000 v Extraction Well PMP12 500 0 5000 32 0 25 1 5000 M Extraction Well PMP13 100 0 5000 32 0 25 1 5000 v Extraction Well PMP14 100 0 5000 32 0 25 1 5000 v Extraction Well PMP15 300 0 5000 32 0 25 1 5000 M Extraction Well PMP16 500 0 5000 32 0 25 1 sood ae X Decision Variable Groups Select fi Decision Variable in Group M Extraction Well Northwest PMP1 PMP2 PMP5 PMPG K Extraction Well Northeast PMP3 PMP4 PMP7 PMP8 K Extraction Well Southwest PMPI PMP10 PMP13 PMP14 v Extraction Well Southeast PMP11 PMP12 PMP15 PMP16 gt gt lt m Ere Next gt Close Help Next to proceed to the next window Defining Constraints The management objectives must be achieved within a set of constraints which may be derived from technical economic legal or political conditions associated with the project There may be constraints on decision variables and state variables and they may take
6. K 1 Output Time 365 00000 day Stress period 1 Time step 1 In the output you will see olive colored cells which represent dry cells First we must switch to the lower layers GoTo Next button to change to Layer2 GoTo Next button to change to Layer3 For this example we are mainly interested in the calculated heads in the excavation location Therefore we can turn off the unnecessary output overlays Viewing Visual MODFLOW Output 21 F9 button Turn on Heads Zone Budget Dry Cells and Pumping Wells Deselect all other overlays OK You may display the exact cell by cell values using the Cell Inspector if you wish To do so amp Tools Cell Inspector from the main menu S Options tab Cell Inspector ox Cell Values Options Grid Position r Row amp Column p Layer CQ Model Position CX World Position CQ Properties CQ Boundary Conditions E Output Head Q Concentration Species Name QV Q Vtotal AllOn All Off Under Grid Position turn on Row Column Layer Then under Output turn on Head amp Cell Values tab Then move your mouse cursor over the excavation zone to see the exact calculated head values You will recall that the head must be lt 29 5 m in order to dewater the excavation site You will see the calculated head values in the excavation zone is 29 30 m 22 Defining MGO Input Cell Inspector rE Cell Values Options Property
7. MGO Tutorial Dewatering Scenario Introduction Introduction Pumping well optimization technology is used to determine the ideal pumping well locations and ideal pumping rates at these locations in order to minimize or maximize a specified criteria Typically pumping well optimization is used in the following applications e Groundwater remediation system design Pumping well optimization is used to estimate the most cost effective pump and treat groundwater remediation system by minimizing factors such as the number of pumping wells and total pumping rates while maintaining capture of the contaminated groundwater plume e Site dewatering system design Pumping well optimization is used to estimate the most cost effective pumping system design by minimizing factors such as the number of pumping wells and total pumping rates while maintaining the required water table drawdown e Water Resources Management Pumping well optimization is used to estimate the maximum yield of a water supply pumping well while maintaining a required water level in the aquifer These are just a few common examples of the many possible applications for pumping well optimization Visual MODFLOW currently supports the public domain version of the Modular Groundwater Optimizer MGO program developed by Dr Chunmiao Zheng and P Patrick Wang from the University of Alabama in cooperation with Groundwater Systems Research Ltd Detailed documentation about the
8. OK to close the dialog Repeat this step for the other injection wells Defining Constraints Se Se Se Add Balance Constraint button located below the table Northeast group at the top INJ2 OK to close the dialog Add Balance Constraint button ca located below the table Southwest group at the top INJ3 OK to close the dialog Add Balance Constraint button located below the table Southeast group at the top INJ4 OK to close the dialog 15 16 Next you must specify the multiplier and constant for the injection wells type 1 for the Multiplier for each well type 0 for the Constant for each well When you are done the Constraints should be similar to the image shown below fl Optimization Options Management Constraints Objectives Control Global Constraints Setting Value gt E Use global rate constraint Yes Minimum total pumping rate 10000 Maximum total pumping rate 0 Ewell location optimization Specify number of active wells Max active wells Max active penalty 500000 State Variable Constraints Select Constraint Type Species Pumping Balance Constarints Northwest INJ2 Northeast INJ3 Southwest INJ4 Southeast oooo State Variable Constraints lt Prev Next gt Close Help The State Variable Constraints are constraints based on system responses to the stresses introduced by the Decisio
9. ONE2 20 2 m 250000 lt Prev Next gt Close C Next to proceed to the next window Defining Optimization Objectives The objectives for an optimization problem can be defined as the net present value of the management costs taken over an engineering planning horizon The costs can include the capital costs associated with well drilling and installation and operation costs associated with pumping and or treatment over the lifetime of the project Other forms of the objective function are also possible For example for a long term contamination containment system the objective function can be defined simply in Defining Optimization Objectives 17 terms of the total pumping volume since the one time drilling and installation costs may be negligible compared to the cumulative pumping and treatment costs The exact form of the objective function depends on the nature of each individual problem In this window define the optimization options Minimize for the Objective Direction For Scenario 1 you will use the Objective s System Installation Cost Total Water Extracted Deselect the other objectives For this objective the default min and max values are sufficient Next to proceed to the next window Defining Control Options The Control Options define the settings that control the numerical aspects of the optimization process These settings are used to tune the optimization process Solver Settings
10. below MGO Tutorial Dewatering Scenario Eile Maps Graphs Tools Row l 600 300 1200 Column J FE zoon EE zoon Fr E8 vert ES puer FIO Hain Layer K E out A Pan ij erag iiy Bl nen General help for this screen For this example we are mainly interested in the calculated heads in the excavation location Therefore we can turn off the unnecessary output overlays e F9 Overlay Deselect all overlays except Heads Zone Budget Dry Cells and Pumping Wells OK You may display the exact cell by cell values using the Cell Inspector if you wish To do so amp Tools Cell Inspector from the main menu amp Options tab Run Unmananged Non optimal Simulation 7 Cell Inspector xi Cell Values Options Q Grid Position Q Model Position World Position Q Properties EC Boundary Conditions R Output RE Concentration Species Name Q Vtotal AllOn Al Off amp Under Output turn on Head amp Cell Values tab Then move your mouse cursor over the excavation zone Zone2 colored dark blue to see the exact calculated head values As you will recall that the head must be lt 29 5 for the excavation area You will see the maximum value is 29 30 m for this model So the current pumping rates satisfy our conditions However in this scenario there are likely too many pumping wells operating 16 resulting in too much water be
11. e active DV For this example you group the extraction wells into their location on the dewatering site To create a new DV Group the Add Decision Variable Group button located below the table and the following Edit Group window will appear Defining MGO Decision Variables and Groups 11 Edit Group ix Group name N orthwest Decision variable type Extraction Well Decision variables in group M D E r A 5 r r r r O F mi mi r Type Northwest at the top for the name Extraction Well for the type amp PMP1 PMP2 PMP5 PMP6 OK to close the dialog The new DV Groups will appear in the table You will now create three more groups using the same procedure Create a group called Northeast and add wells PMP3 PMP4 PMP7 PMP8 Create a group called Southwest and add wells PMP9 PMP10 PMP13 PMP14 Create a group called Southeast and add wells PMP11 PMP12 PMP15 PMP16 This concludes the requirements for Decision variables If you have entered the data successfully your window should be similar to the figure shown below 12 Defining MGO Input fl Optimization Options Management Constraints Objectives Control Optimization Engine MGO Modular Groundwater Optimizer gt Decision Variables Select Variable Type N Initial Rate m 3 d Min Rate m 3 d Max Rate m 3 d Rate Steps NPStep M Extraction Well PMP1 100 0 5000 32 0
12. ing extracted from the system which can become very costly MGO Tutorial Dewatering Scenario You will now use MGO to optimize this scenario MGO will help determine a pumping strategy that is needed to reach these head constraints and minimize costs this includes an optimal pumping rate and an optimal number of pumping wells that should be installed We will now proceed to define the MGO optimization settings Defining MGO Input To load the MGO input module Se File Main Menu from the top menu bar Run from the top menu bar amp Optimization Well Optimization from the main menu This will load the Optimization Options window as shown below My Optimization Options Management Constraints Objectives Control Optimization Engine MGO Modular Groundwater Optimizer gt Decision Variables Initial Rate Min Rate m 3 d Max Rate m 3 d Rate Steps NPStep m 3 d Variable Type Decision Variable in Group MGO for the Optimization engine You will now define the decision variables for the dewatering problem Run Unmananged Non optimal Simulation Defining MGO Decision Variables and Groups A Decision Variable DV is a model input that may be influenced or controlled by the user These are the inputs that you are trying to optimize For MGO the DVs are limited to Pumping Wells Extraction and or Injection Wells Adding Decision Variables In thi
13. n Variables These typically include constraints such as minimum and maximum allowed values of head or concentration at a selected location For this simulation we will add the Head constraint which is the water elevation we are trying to achieve in order to do the excavation Add State Variable button HF located below the grid In the Add State Variables Dialog amp Head under State Variable Type Defining MGO Input s Zone2 which contains the excavation zone OK In the State Variables Constraints grid define the bounds for the constraints type 20 m for the Lower Bound type 29 5 m for the Upper Bound type 250 000 for the Penalty value Once you are finished the Constraints window should be similar to the one shown below E Management Constraints Objectives Control l Global Constraints Pumping Balance Constraints Setting Select Unmanaged Well Decision Variable Multiplier Constant gt E Use global rate constraint Group i Minimum total pumping rate b M Ni Northwest 1 0 Maximum total pumping rate 1 E10 M N32 Northeast 4 0 E Well location optimization Specify number of a M N3 Southwest 1 0 Max active wells M INI Southeast 1 0 Max active penalty 500000 Whether to use global rate constraint State Variable Constraints Select Constraint Type Species Location Type Location LowerBound Upper Bound _ Units Penalty b M Head N A Zone Z
14. ost per well 1 000 Drilling Installation cost per unit depth of well 0 000 Pumping Treatment cost per unie vol une of flow 1 000 Mass removal cost factor 0 External objective function muleiplier 0 000 Number of Optimization Parameters 16 Parameter Scaling Factor 1 pes Number of Parameters Active at Any T Penalty if Active Parameters More than specified Maximum 500000 ParamNo Mimimum Maximum of Discret K SP1 SP2 Depth ioptonoff start Do you want to import Decision Variables into Visual MODFLOW Import Finally the MGO Output file MGO displays the MGO simulation input file and output file Defining MGO Input Now we need to confirm that these optimized injection extraction rates have satisfied the constraints we have defined To do so we must view the Visual MODFLOW results Since we do not want to overwrite the original pumping rates with these calculated pumping rates we will not import the MGO results into Visual MODFLOW Close button Next we will load the output of Visual MODFLOW Please continue in the next section Viewing Visual MODFLOW Output To load the Visual MODFLOW output Output from the main menu bar The Visual MODFLOW output window will load File Maps Graphs Tools Help View Column View Row View Layer Head 7 Options Time Select Export Layer gt Zz Row 1 Column 3 Layer
15. raction We force the total sum of extraction rates to equal the total injection rate because all water obtained from extraction wells will be reinjected into the ground This is often important for water supply environmental and economic reasons In MGO Injection wells will be added in the constraints as unmananged wells Unmanaged wells are pumping blocks whose rate is fixed in the input and are not used as a decision variable or computed in the MGO simulation We will add the four injection wells in order to balance out the flow of the four extraction wells Each injection well will have a 1 0 multiplication factor to balance the rates We will assign one injection well to each group of extraction wells In order to specify a Balance Constraint for MGO using Visual MODFLOW the unmanaged wells must already be present in the model and the managed pumping wells may only be selected if they belong to a DV Group You will recall that we have already added this DV group in the Management tab To add a new Balance Constraint Add Balance Constraint button located below the table The following Add Balance Constraint s window will appear Defining MGO Input Se Add Balance Constraint s Well Group Northwest gt Unmanaged Wells Type Injection Well Injection Well Injection Well Injection Well Northwest group at the top Next you must select one unmanaged injection well Se Se INJ1
16. re indicated press the lt Tab gt key a press the lt Enter gt key click the left mouse button where indicated eS double click the left mouse button where indicated The bold faced type indicates menu or window items to click on or values to type in denotes a button to click on either in a window or in the side or bottom menu bars Opening the Visual MODFLOW Model Getting Started On your Windows desktop you will see an icon for Visual MODFLOW Mee ae Visual MODFLOW to start the program The Dewater model has already been built for you to open this model 2 MGO Tutorial Dewatering Scenario File Open from the top menu bar Browse to the Tutorial MGO folder and locate the Dewater VMF file Select this file and es Open You will now briefly examine the Visual MODFLOW model Input from the top menu bar This will load the input window of Visual MODFLOW As you can see the model domain is 20 rows by 20 columns by 5 layers The X extent is 2000 m Y extent is 2000m Total area is 988 acres We will now briefly review the inputs for this model Wells Pumping Wells from the top menu bar to see the well data The study area has 16 candidate extraction wells PMP1 PMP16 screened over the bottom two geological layers and 4 injection wells INJ1 INJ4 screened over layers 3 and 4 The initial rates of these wells are as follows E
17. s exercise you will add all the candidate extraction wells as decision variables the Add Decision Variable button located below the table and the following Add Decision Variable s window will appear Add Decision Variable s Decision variable type Extraction Well Select decision variable s IMMA This window lists all of the Extraction Wells and or Injection Wells in the model Extraction Well at the top Select all the wells from the list OK to close the dialog For each decision variable enter the following Min Rate type 0 m3 day For each decision variable enter the following Max Rate type 5000 m3 day 10 Defining MGO Input Note In the Well Optimization packages pumping rates are treated as absolute values It is not necessary to define extraction rates as negative values or injection rates as positive values For each decision variable enter the following Water Cost type 0 25 For each decision variable enter the following Mass Cost type 1 For each decision variable enter the following Install Cost type 5000 Defining Decision Variable Groups A Decision Variable Group DV Group is a collection of individual DVs which may have similar interests These DV Groups are used for different purposes depending on which optimization engine is selected When using MGO a DV Group may only be used to define pumping balance constraints A DV Group must contain a minimum of on
18. the form of either equalities or inequalities Constraints on the decision variables include the number of candidate wells the upper and lower bounds for pumping injection rates and the candidate well locations Constraints on the state variables might include the requirement that hydraulic heads be maintained above or below a certain level or that contaminant concentrations not exceed regulatory standards at specified compliance points Zheng and Wang 2003 In this example we will define constraints on the pumping rates and on the head values for the pumping well screens Defining Constraints 13 Global Constraints The Global Constraints are applied to all Decision Variables as a whole and influence the total pumping rate and the total number of active pumping wells Under the Global Constraints define the following options Yes to Use Global Rate Constraint type 10 000 for the Minimum Total Pumping Rate value type 0 for the Maximum Total Pumping Rate amp Specify number of active wells for the Well Location Optimization type 4 for the Max active wells type 500 000 for the Max active penalty Pumping Balance Constraints The Pumping Balance Constraints establish the dependence relationships among different wells They are imposed to link the flow rates of unmanaged wells with those of managed wells This option is only available for the MGO engine In this scenario we want to specify that Total Injection Total Ext
19. timal extraction rate for three or four extraction wells It also calculated an equal injection rate for three or four injection wells such that the constraint that injection extraction is satisfied In addition the global constraint of 10 000 m3 day for maximum pumping rate was not exceeded Optimization Results x Decision Variables Objective Function Dewater mgo Obj Value 1723060 6 2223060 6 723060 6 1223060 6 1 6 11 16 iterations Do you want to import Decision Variables into Visual MODFLOW Import In the Objective Function tab we can see progress of the calculated objective function value through the course of the simulation at each iteration You will recall for this simulation we used 20 iterations Optimization Results ms Decision Variables Objective Function Dewater mgo aeaass Fz m GO wF Project Dewater Optimization module Input File For MGO E r2003 The file is translated at Fri Mar 11 15 56 47 2005 Optimization Solver Selected Genetic Algorithms Response Database Not Save Maximum Number of Teerations Number of Forward Simulations per Iteration 100 Number of Iterations for Convergence 100 Restart Option Not Activate Debugging Level O Intermediate Simulations Not Saved Final Simulation with Optimized Parameters Required Direction of Optimization gt Minimization Coefficients for the Objective Function Fixed capital c
20. xcavation area This zone is colored in dark blue MGO Tutorial Dewatering Scenario Zone 2 will contain the head constraints The goal is to lower the head to 29 5 m so that the top two layers can be excavated without becoming saturated Run Unmananged Non optimal Simulation You will now run the model to see the results of the dewatering using the current pumping rates without optimization File Main Menu from the top menu bar Run from the top menu bar Run from the main menu bar a second time Translate amp Run The MODFLOW 96 numeric engines will then run This should take approximately 30 seconds depending on system resources When this is complete Close button Then to see the output Output from the main menu bar GoTo Type 1 F9 Overlay amp Check the box beside C I Contour labels The output window will appear similar to the figure shown below Run Unmananged Non optimal Simulation 5 File Maps Graphs Tools z Row I Column J Layer K 1 Output Time 1 00000 day Stress period 1 Time step 1 The olive colored cells represent dry cells in the upper layer The upper layer is nearly completely dry due to intensive pumping Next we need to move to layer 3 which contains the zone for the head constraints GoTo Next button to change to Layer2 GoTo Next button to change to Layer3 and you should see the contoured heads as shown
21. xtraction Wells Rate m3 day PMP1 100 PMP2 100 PMP3 300 PMP4 500 PMP5 100 PMP6 1500 PMP7 1500 PMP8 500 PMP9 100 PMP10 1500 PMP11 1500 PMP 12 500 PMP13 100 Opening the Visual MODFLOW Model 3 Extraction Wells Rate m3 day PMP 14 100 PMP15 300 PMP 16 500 The initial rates for the injection wells are as follows Injection Wells Rate m3 day INJ1 100 INJ2 100 INJ3 100 INJ4 100 r Se ee Properties Conductivity from the top menu bar and then Database button from the side menu to see the conductivity zones for the model The default conductivity for this model is 8 64 m day and 0 864 m day vertical conductivity Boundaries General Head from the top menu bar and then Edit gt Group from the side menu to see the boundary conditions for the model There is a general head of 43 m on the west boundary and 47 m on the east boundary ZBud from the top menu bar and then Database button from the side menu to see the defined zones for the model The excavation zone is in Layer 3 So we must switch to the lower layers GoTo Next button to change to Layer2 GoTo Next button to change to Layer3 The Wells overlay obscures the excavation area zone To correct this ee F9 Overlay button In the Overlay dialogue deselect BC F Wells overlay Se OK This model contains two zones e Zone 1 default entire model domain e Zone 2 e
Download Pdf Manuals
Related Search