Office of Solid Waste and
Emergency Response
(5102(3)
EPA 542-F-04-002
February 2004
clu-in.org/optimization
Improving Pumping Strategies for Pump and Treat Systems
With Numerical Simulation-Optimization Techniques
Demonstration Projects and Related Websites
Why Optimize? To identify pumping strategies that:
minimize life-cycle costs, annual costs, or cleanup time while assuring protectiveness
maximize mass removal
minimize pumping rate required for plume capture
What are Simulation-Optimization Techniques? These are mathematical techniques that
couple simulations of groundwater flow (e.g., MODFLOW) and/or contaminant transport (e.g., MT3D) with
mathematical optimization algorithms, to determine an optimal solution when many possible solutions exist.
Bad solution
Potential solution
Potential solution
Potential solution
Potential solution . Good / ution
Bad solution
Potential solatia
Very good solution
Very good solutigj
Bad solution
Potential solution
Very good solution Bad sd faon \J Potential solution
Good solution Potential solution., . V , . Bad solution
Potential solution
E
OPTIMAL
OLUTION
Good solution
Bad solution
Bad solution
Potential solution
iry good solution
Bad solution
Good solution
Very good solution
Potential solution
Very good solution
Bad solution
The simulation-optimization approach is more efficient than simulating a small number of pumping scenarios in
a "trial and error" manner (the traditional approach), and typically yields a much improved result. There are
two general subclasses of simulation-optimization methods for ground water pump and treat systems:
Hydraulic Optimization - based on ground water flow modeling (not transport), most appropriate when
hydraulic containment is the primary concern
Transport Optimization - based on groundwater flow and transport modeling, most appropriate when
aquifer restoration is the primary concern (containment can also be considered)
Preliminary Steps
Optimization Formulation
gio
W
ca
V) QO
• 01 fs*
Environmental Protection Agency
>n 5, Library (PL-12J)
est Jackson Boulevard, 12th Fid
go, IL 60604-3590
Consider
system goals
Create
numerical
model(s)
suitable for
design
purposes*
h
*Hydraulic optimization
flow and contaminant ti
Determine an appropriate objective that can be
represented by a mathematical function and
can be minimized or maximized
Determine constraints (limits on well locations,
pumping rates, water levels, capital costs, etc.)
Select a suitable simulation-optimization code
*Hydraulic optimization only requires a groundwater flow model. Transport optimization requires both groundwater
flow and contaminant transport models. "Optimize" involves many systematic runs of the simulation model.
-------
What demonstration projects have been completed? EPA and DOD have highlighted two
projects that demonstrate the application of simulation-optimization techniques for pump and treat systems at
multiple real-world sites.
Applicable Sites
Type of Ground Water
Model
Optimization Technique
Agencies Funding the
Demonstration
Demonstration
Completion Date
Number of
Demonstration Sites
Summary of Results
Hydraulic Optimization
Demonstration Project
Priority is plume containment
Ground water flow model (e.g., MODFLOW)
Linear and mixed-integer programming
USEPA Technology Innovation Office
(ow known as the Office of Superfund
Remediation and Technology Innovation, or
OSRTI)
1999
3 sites (multiple formulations per site)
Potential cost avoidance of millions of dollars
in life-cycle costs were demonstrated at two
of the three demonstration sites
Transport Optimization
Demonstration Project
Priority is plume cleanup or mass removal,
containment can also be considered
Ground water flow and transport model (e.g.,
MODFLOW/MT3DMS)
Nonlinear programming using global search
algorithms
Department of Defense ESTCP Program
(lead agency was the Naval Facilities
Engineering Command) and USEPA
Technology Innovation Office
2003
3 sites (three formulations per site)
In every case, both simulation-optimization
codes outperformed a third group using
traditional "trial-and-error" simulations,
representative improvement was 20 percent
improved pumping strategies could yield
$ millions
in life-cycle cost savings at some sites
Lessons Learned
Simulation-optimization approaches
can be effectively applied at real-world sites
A 20 percent improvement in the objective
function value (the item being minimized
or maximized) is typical
Optimization results can not be generalized
a specific analysis for each site is required
Final reports and details pertaining to each demonstration project (including codes), and general information
on simulation-optimization techniques, can be found on the Federal Remediation Technologies Roundtable
(FRTR) website listed below.
Federal Remediation Technologies Roundtable Website (http://www.frtr.gov)
Hydraulic Optimization Demonstration Project
http:llwww.frtr.govloptimizationlsimulationlhydrauliclgeneral.html
Transport Optimization Demonstration Project
http:llwww.frtr.govloptimizationlsimulationltransportlgeneral.html
General Information
http:Hwww.frtr.govloptimizationlsimulation.htm
USEPA Contact: Kathy Yager (OSRTI), 703-603-9910, yager.kathleen@epa.gov
-------
|