Department of Mathematics

QED Environmental Systems Inc.

Proposed Project for the MSU Industrial Math Students


Active and inactive domestic solid waste landfills (US) require the installation of systems to control and capture methane produced by the anaerobic processes breaking down the waste.  Safety, green house gas control (methane has 25x the green house gas impact of CO2) and capture for energy use are all reasons these systems are installed.  Additionally, regulations govern the methane gas emissions for air pollution and safety reasons.  While some landfills employ passive flare systems to burn off the methane, this study will focus on larger, controlled systems that use a large vacuum pump, piping and wells to draw the methane to a central location at the site for use generating electricity or gas recovery.

Data is currently collected from the numerous extraction points (gas well heads), often as frequently as once a week.  A typical landfill may consist of 50 to several hundred well heads.  The frequency of data collection, combined with the number of data collection points, creates a large, multivariate data set that grows over time.  This data (landfill gas data – LFG data) is typically stored in spreadsheets or enterprise level environmental databases, but current data use is restricted to flagging exceedences, some charting and (more rarely) some data trend mapping as related to GPS location of the well point.

Description of Data Sets:

  1. Currently collected data - data is currently collected by field technicians that visit the well heads and take data on a number of parameters, an example of which is shown in the table below.  Issues of field instrument accuracy, precision and calibration are not currently integrated into the data set.
  2. Static site/position data - there will also be a data set of various fixed parameters relating to the data collection point (well head), such as GPS location (formats like GPS-UTM –, well depth, elevation, screen length, etc.
  3. Other available data – other, time based data may be useful, such as barometric pressure (and other weather and temperature data) taken from nearby Weather Underground stations.
  4. Future data additions – certain data inputs that do not exist today may be useful from a process control standpoint, such as sensor based vacuum or temperature data (to increase data resolution beyond weekly readings), control valve position as a repeatable number, etc.

Problem 1: Multivariate Analysis of Large Data Sets from Landfill Gas Colletion Systems

Use statistical/Taguchi methods to discover data relationships – this would be an effort to dig deeper than the current single variable data analysis and contour mapping of individual data parameters that has gone on in the past.  Are there relationships between parameters and/or other drivers, such as barometric pressure that explain data trends? Can analysis of data trends create information about system “noise” or field technician sampling variations? Further, can this work suggest changes in how data is collected or analyzed to give a better picture of the overall multi-well system and the controls that are used to manage it?  Does this work result in a site-specific numerical model that can be used for predictive operational control?

Problem 2: Developing Predictive Models of Landfill Gas Data Sets for Process Management

Use PID control theory to predict system changes and drive convergence – can classic PID control techniques be employed to give field technicians recommended operational control changes that result in more stable overall systems?  This assumes that current data/control changes are driving systems where hysteresis is the result of non-precise, or incorrect control changes.

Problem 3: Utilizing Data Visualization Tools to Deepen the Understanding of Large Landfill Gas Data Collection

Use 3D and time based data visualization tools to discover hidden data relationships.  This is related to #1. – can a presentation similar to this example –  2011 Japan earthquake data visualization – – help deliver a deeper understanding of system data variations as they relate to both control changes and external drivers (such as barometric pressure, air temperature, etc.).

Top of Page