TODO.txt

verify np 0.8.5
    verify metric model
        write unit tests
            test demographic growth
                make sure that population does not change when the population growth rate is zero
                make sure that population shrinks when the population growth rate is negative

            test economic growth
                assert that the demand multiplier is one when the economic growth rate is zero
                assert that the demand multiplier is one when the elasticity is zero

            zero household demand
                expect the low voltage cost per household to be zero when the household demand is zero

            zero other demands
                see what happens when there is no commercial or productive demand
                expect all costs to be zero and no technologies chosen when all demands are zero

            test costs
                check LV/HH value
                expect capital costs to be zero when component costs are zero
                expect recurring costs to equal total cost when capital costs are zero
                expect capital costs to equal total cost when recurring costs are zero

            test technology selection
                expect that a node will have the technology with the least overall cost
                expect the number of off-grid nodes to increase when facility demand is zeroed

        write parameter tests
            expect not to have diesel generator costs if there is no commercial or productive demand
            expect the net recurring costs to increase when the time horizon increases
            expect the net recurring costs to increase when the interest rate increases
            expect the demand to increase when the economic growth rate increases
            expect the demand to increase when the elasticity increases
            expect the mv cost per household to decrease when demand increases (economy of scale)
            expect the lv cost per household to decrease when demand increases (economy of scale)
            expect more nodes as grid when demand increases
            expect the number of households to increase when the population increases
            expect the demand to increase when the population increases
            expect the # of urban households to decrease when the urban household size increases
            expect the # of rural households to decrease when the rural household size increases
            expect the # of urban households to decrease when the urban threshold increases
            expect the LV cost to increase when the mean interhousehold distance increases
            expect the # of nodes with grid to increase when the base household and productive demands increase
            expect the # of nodes with off-grid to decrease when the base household and productive demands decrease
            check demand curves by setting all demand curve points to y=1 and check that all households have base demand
            expect the peak demand to increase when the peak fraction increases
            expect the peak demand kW to decrease when the annual peak usage increases but demand kW-h/year stays constant
                note that usually demand increases when annual peak usage increases

            make sure that the system always picks a system with larger capacity than the required demand
            expect to have fewer of a given technology when that technology is made more expensive
            expect to have increased replacement costs when increasing lifetimes
            expect to have increased recurring costs when increasing OM or fuel costs

release np 0.9.2
    improve output
        put attributes in shapefile

    improve documentation
        consider adding hint bubble + spotlight tutorial
        add one glimpse overview to NP homepage
        add indices to NP documentation
        add my description of peak nodal demand

    deploy
        implement fabric SSH deployment script (thanks to Alex Dorey)

    improve metric model
        think how to estimate demand if facility count is less than one
        consider making MID different for rural vs urban
        consider making number of peak hours different for rural vs urban
        make it possible to specify electricity costs in usage brackets
        consider adding support for exchange rate
        consider replacing scaling factors and base unit demand with direct demands
        consider using Decimal class for more accuracy

    improve architecture
        consider applying surface maps of cost
        consider applying surface maps of demand
        consider putting modelling blocks into modules
        consider making it possible to read a metric model from a spreadsheet

    improve network algorithm
        fix network algorithm to handle case when new grid affects results (might require iteration)
        remove geometric intersection check for nodes because it doesn't work for real data
        consider using heapq to get smallest segment
        consider using roads between electrified communities to approximate location of existing grid line

    improve user interface
        fix column headers and filter in dataTable to be always visible even when scrolling
        use Layer.getDataExtent instead of computing bounds
        consider changing wording of commercial facility to "vendor"
        add fallback WMS layer in case there is no internet
        make it possible to see map on whole screen
        let user click on difference for system count to see which communities are different
        increase scenario name length
            add error message in controller if scenario name is too long

        make it clear that base unit demand corresponds to scaling factor = 1
        revise productive unit demand to be per household per year
        enable node-level overrides directly in page
        make it easy to see tabulated costs
        specify projected per node where applicable

    fix bugs
        restore hover after scope selection update
        check why we cannot specify TimeHorizon=1

    improve usability
        make it easier to use default datasets
            add datasets at three scales (local, national, continental)

        change input form to use tabs
        enable links to default templates to be cloned

release np 0.9.3
    enhance user interface
        fix node detail
            * fix job view so that it can handle foreign characters using unicode
            enable comparison between node detail variables
                display node details for comparison scenario as well

        improve map interface
            add measurement tool
            enable toggling on/off for Google map labels

        fix scenario creation
            add quick plot for curves
                add instant ajax curve plots

            handle case if scenario name is too long

        handle unicode in scenario name
        enable user to compare custom community-level columns in the summary table
        consider delaying login until after the user tries to submit a new scenario
        move social infrastructure count from grid section to demand section

    put duplicate nodes warning in warnings output
    consider reducing package size by storing and sending parameters and outputs as lists instead of dictionaries
    figure out some way to ensure that large jobs are run on the cluster machines with more horsepower
    put large jobs on separate queue
    consider reimplementing search radius, but determine it automatically
    save segments to sqlite temporary file during segment generation phase
    consider weighting regions by population density when computing in zoomed out scale
    change beaker session secret
    figure out why sometimes rabbitMQ queues disappear
    update metric estimation models
        consider having a connection cost for mini-grid just like for grid (although i recall Aly wanted to remove it for mini-grid)

    update network optimization models
        add 4th option to consider optimality of standalone subgroups connected to each other
        consider having a search radius that is different for each node, computed from local metrics
            set search radius to be (current subnetwork's mvMax) + (local maximum mvMax of all other subnetworks)

release np 1.4.0
    fix network building algorithm
        fix minimum subnetwork size so it does not eliminate nodes on existing grid
        enable large distance connections
            consider how to add HV
            consider when each node has limitations on what technologies are available

        think how the existence of nodes with existing grid will affect the network building algorithm

    update metric models
        add warning if largest population requires extrapolation because it is outside the point range for the curves
        figure out how to pay attention to whether a node is already electrified
            modify nodal costs so that they take into account whether the place has existing grid or not

        consider adding civil engineering cost (constructing building to house generator) as a separate cost to mini-grid based on % gen
        implement simplified X,Y,demand metric model
            consider adding simplified mvMax model with initial + maintenance + fuel

        add pure PV as tech option
        add wind as tech option
        rethink how to do grid internal system
            change gridInternalCost so that communitys with large spatial area have extra MV cost (look in kenya paper)
                place transformers first based on community area
                figure out length of extra internal mv line

        consider how to implement MV reinforcement cost to gridExternalCost on MV line near grid extension
        add way to mark whether a node has existing grid, mini-grid, off-grid

    enhance user interface
        enable measuring distance between nodes and existing grid
        add segment weights to global outputs csv
        use openlayerer to hand select the parts we need
        group datasets into locations
            show locations via polygons

        consider adding tooltips for fields
        consider adding self-served WMS backup

release network-planner 1.5.0
    increase types of input that the system can accept
        let the user enter nodes from a map

    add feedback after user uploads demographic data
        map communitys in demographic layer
            add feedback when user clicks on a node
                plot rural and urban populations over time horizon for a single node

        map administrative regions (i.e. states or districts) in an administrative layer of the map.
            add feedback when user selects administrative region from a drop down
                show demographic statistics for the selected administrative region

            map resource availability by region
            map resource cost by region

        add feedback when user enters curve points
            plot curve immediately so that user can verify the shape of the curve

        add feedback after user enters metric parameters
            display a pie chart of demands aggregated over all nodes
            display a demand layer that shows the community nodes sized differentially according to the each node's demand
            display an administrative demand layer that colors each region differently according to the aggregated demand for the region
                visualize demand via colormap as they are aggregated by administrative level

            show where each population was interpolated on demand curve plots
            show where each population was interpolated on count curve plots
            generate three graphs of peak demand by population for household, institution and total peak demand
            add feedback when user clicks on a node
                show pie chart break of the demands for that node

            add feedback when user selects an administrative region from the dropdown
                show a pie chart breakdown of the demands for the selected administrative region

        add feedback after user enters cost parameters
            plot a histogram of system cost vs. system type aggregated over all nodes
            add a new cost map node layer to size each node differentially based on cost by diesel
            add a new cost map node layer to size each node differentially based on cost by photovoltaic diesel
            add a new cost map node layer to size each node differentially based on cost by grid
            add a new cost map node layer to show a different icon for each node depending on best technology choice
            add a new administrative cost map layer to show each region in a different color based on cost by diesel
            add a new administrative cost map layer to show each region in a different color based on cost by pvdiesel
            add feedback after user clicks on a node
                show histogram of system cost vs system type for the node
                show mvMax for the node
                    visualize mvMax

            add a new administrative cost map layer to show each region in a different color based on cost by grid
            add a new administrative cost map layer to show each region in a different color based on best system type
            show a plot showing the number of nodes binned by least cost system type

        add feedback after user asks application to build network based on metrics
            display existing grid on map in existing grid layer
            display proposed grid extension on map in proposed grid layer

        add feedback after user asks application to generate feasible rollout plan based on computed network
            show a movie of the rollout plan
            show plot of how each type of demand varies over time, aggregated over all nodes
            show a plot of how each type of cost various over time, aggregated over all nodes
            add feedback after user clicks on a node
                show a list of what will happen to the node during the course of the rollout grouped by year
                show how the demand of the node will vary during the course of the rollout grouped by year
                show a list of what the node should expect to pay during the course of the rollout grouped by year

        show information aggregated over all nodes
            plot rural and urban populations over time horizon aggregated for all nodes

    update metric model
        consider how to model the ability to pay
        vary the engine cost per kilowatt based on a linear or logistic function (economy of scale)
            make cost of electricity decrease for larger generators
            map engine cost directly to diesel system cost using a logistic curve

        use projected population at specific points in time
        set same default threshold populations for Count Curve Points and Demand Curve Points
        scale base unit demand to minimum threshold population point
            indicate that the value of the population corresponding to 1 is the most accurately represented by a curve

        distinguish between peak demand and peak capacity
        add target electrification rate (how is this different from penetration rate?)
        add time varying growth rates by enabling rates to be entered as list
        add aggregated outputs to job output
        split off-grid demands to size panels separately for each type of demand

    update network sequencing algorithms
        implement basic rollout code
            make sure capital costs are discounted by the year they are constructed
            make sure recurrent costs are discounted after capital costs

        consider three stage network optimization (with HV for large cities first, then MV, then LV) depending on source prices

    update user interface
        give live feedback during processing
        group results by extra attributes in each node
            make it possible to group results by administrative level like state
                aggregate cost, growth, demand by region

            aggregate by different attributes

        record elapsed computational time
        show custom satellite image backdrop
        add aggregated cost breakdown for grid
        let user pick what columns are output in csv
        enable comparing nodes

    update tests
        add unit test to ensure that callbacks work properly

    add plain language description for each variable


release network-planner 2.0.0
    update data import modules
        consider dealing with CAD files

    update metric estimation models
        make it easy to add another technology option
        expand software for use in water pipe design
        include topography such as mountains, valleys, obstacles
            let the user specify no-go-through zones or preferable zones

        consider how to optimize load flow
        consider adding industrial demand like factories
        consider adding institutional demand like government buildings
        modify cost code to support increasing electricity access in a partially electrified community i.e. intensification
        determine high-voltage line cost
            determine which lines should be high-voltage based on number of people served by that line (thanks to Alex Hofmann)

    update network building algorithms
        redo network modelling to follow roads
        think of how to model the sources of energy as well as the sinks (demands)
        recompute costs for a network based on wire type (HV, MV, LV)
        implement neuron growth model like the peripheral nervous system
        implement slime mold model
        consider adding flow models that depend on time based on cost efficiencies
        ask selin for new network code
        email selin and aly the new optimization problem description
        * consider performing optimization in iterations because a new grid might appear and change optimality of grid for certain nodes
        fix wire cost so that it depends on distance and load (how much electricity it is carrying)
        consider enabling iterative equilibrium computation for cost/demand using game theory
        consider enabling connection of regions, perhaps weighted, to reflect borders and density

    improve user interface
        add email notification for completed jobs
        show flowchart of model as moving along

    fix realistics
        make it more expensive to go through an urban area than a rural area
        make cost iteratively interlinked with demand because if there is more demand, it will affect cost and vice versa

    update data export modules
        investigate how to create interactive maps in PDF for download