WashOpt User’s Guide


Welcome to the WashOpt user’s guide! WashOpt combines operations research tools and GIS to help you generate novel water and sanitation infrastructure development scenarios. WashOpt is under steady development to provide users the capability to rapidly evaluate novel combinations of WASH technologies under wide varieties of performance expectations, and to generate development plans that hedge against the performance uncertainties of water and sanitation treatment technologies.

Tools

Navigate Map: Because WASH4All is focused on water treatment in Honduras currently, the map centers on Honduras by default. You can pan the map by clicking and dragging it, much as you would in Google Maps. You can zoom the map in three ways:

1) Double-click at a point to zoom in there
2) Roll the mouse wheel forwards to zoom in at a point, or backwards to zoom out
3) Click the + or – buttons in the top left corner to zoom in or out

Draw Boundary: WashOpt takes the approach that to generate a development scenario, you need to know at least where you’re working, what need you seek to address, and what options you have to address that need. To set the area of your scenario, click “Draw Boundary” on the WashOpt Control Panel, and then draw a rectangular boundary by left-clicking the mouse (hold the button down) and drawing two diagonally opposed corners of a rectangle.

For the purposes of this demo site, the other two questions are answered for you: Your scenario will address waterborne disease burden (measured in terms of diarrheal prevalence), and to address this need you will construct hypothetical interventions — point-of-use treatment systems (each with the capacity to treat drinking water for one person daily) or large-scale water kiosks (five different types, with the capacity to treat drinking water for 1000, 2000, 3000, 5000, and 10000 people daily, respectively). We are excited at this first step, and work is currently underway to include piped drinking water distribution networks and sanitation interventions in WashOpt.

Run Scenarios

Once your scenario is designed, click the “Run scenarios!” button. This will initiate a call to the server, which will then construct an optimization file for the scenario, find the optimal solution for the problem posed by your scenario, read the optimization output file, build a GIS representation of the infrastructure elements in the output file, and prepare these elements for display on the webpage.

Interpretting Scenario Output

After processing, WashOpt will present solutions for nine scenarios, detailed below by their number, and the assumed effectiveness (low, average, or high) or the POU and kiosk interventions:

Scenario 1: Point-of-use (POU) solutions give average performance, and kiosks give average performance
Scenario 2: POU average, kiosk high
Scenario 3: POU average, kiosk low
Scenario 4: POU high, kiosk high
Scenario 5: POU high, kiosk average
Scenario 6: POU high, kiosk low
Scenario 7: POU low, kiosk low
Scenario 8: POU low, kiosk average
Scenario 9: POU low, kiosk high

Additionally, if you selected any of the hedging algorithms, WashOpt will present hedged solutions at N equally spaced budget intervals from the stated budget down to $0. Currently N is set at 2.

Scenario updating: When you generate a scenario (or set of scenarios) by pressing the Optimize button, the webpage keeps track of your bounding box. If you wish to adjust an option (like lowering the budget), you can hit Optimize again without having to respecify the geographical boundary.

Intermediate Options: Budget

Next you’ll want to set a budget for your scenario. Infrastructure costs are purposefully unscaled currently, and are outlined below:

*Point-of-use (POU) unit: $0.002
*1000-person capacity water kiosk: $1
*2000-person capacity water kiosk: $1.5
*3000-person capacity water kiosk: $1.7
*5000-person capacity water kiosk: $2
*10000-person capacity water kiosk: $3

Once you have picked a budget, enter the number in the “Budget” text field on the Optimization Controls panel. If this is your first time to use WashOpt, consider skipping the Advanced Options section for now, and proceeding to Optimize.

Advanced Options: Treatment effectiveness

Background: WashOpt bases its estimates of population distribution and waterborne disease burden on publicly available datasets and citizen-submitted data. We are currently building a small-scale distribution estimate of waterborne disease burden, so for now the program naively assumes that uniformly across Honduras, 10% of the population is suffering from diarrhea due to waterborne disease at any time. It is further assumed that consuming of water treated by POU units and kiosks (instead of consuming water of unknown treatment status), will reduce the prevalence of diarrhea by some percentage.

These are, loosely speaking, reasonable types of assumptions to make, but attaching reliable numbers to such assumptions is difficult. The uncertainty of water treatment effectiveness is a major area of public health and engineering research (and a primary motivator for the development of the WashOpt tool). Currently, WashOpt uses only low and high estimates of treatment technology effectiveness, and imputes a middle value halfway between the two.

Advanced options: Hedging

WashOpt attempts to address the uncertainties in the data in two ways: (1) evaluating the full set of scenarios that result from varying the POU and kiosk effectiveness estimates independently, and (2) three hedging strategies. The hedging strategies incorporate information from the basic scenario generation, and use different algorithms to create a novel scenario that balances performance with respect to all scenarios generated. So for example, in basic scenario generation you create solutions for the scenarios that POU treatment is more effective than kiosk treatment, and vice versa, and that they are equally effective, because you don’t know for certain (there’s just not enough data out there yet). The output of each scenario is the best development solution under the given scenario, but who can say which scenario, in practice, is actually more likely? WashOpt uses hedging algorithms to create a solutions that perform reasonably well under a wide variety of scenarios. This is akin to the investing concept of diversifying to reduce risk. Specifics of the WashOpt hedging algorithms will be made publicly available in the next few months. In the meantime, you can employ these hedging algorithms (named V2, V3, and V4) by clicking on the checkboxes of those names on the Optimization Controls panel.

*Please note that the hedging operations can currently take a long time to complete (on the order of 2-10 minutes). We are implementing several approach to improve this performance.