Help: THERMOS network model user interface

There are a few single-key shortcuts in the map view:

The map view is split into three regions: a map, the selection info panel on the right, and the network candidates panel at the bottom.

You can change the size of the panels by clicking and dragging on the grey border between them.

You can quickly hide or show a panel by clicking on the little arrow button in the middle of the grey border: .

The main purpose of the map view is to select candidates (these are buildings or pipe routes), and then to manipulate or investigate them. The selection refers to the set of candidates which are selected at any time.

The selection is ephemeral and only one selection exists at a time, much like a selection in a spreadsheet program. There is no quick way to save or recall a selection, so if you want to do something very complicated it is best to do it a bit at a time.

You can see on the map whether something is selected by the style with which it is drawn. Selected buildings are drawn with a darker colour, and selected paths or buildings are drawn with a thicker line.

Clicking on a thing with the mouse is the simplest way to select it. This will also deselect whatever is currently selected. Clicking in an empty space will therefore clear the selection.

If you want to select several things by clicking on them, you can use modifier keys to change what happens. Holding down a modifier key while clicking will control how the selection is affected:

Shift+click

Holding the Shift key while clicking will add whatever is clicked to the selection, without unselecting what was selected before.

Control+click

Holding the Control key while clicking will invert membership of the selection, without changing the rest of the selection. For example if you Control-click something that is in the selection, that thing will be deselected. If you control click something which isn't selected, it will be added to the selection.

There are other ways to control the selection, which we will come onto in:

• Area selection
• Narrowing the selection by category
• Selection column

Around the edge of the map are some control buttons:

The selection info panel is the region on the right hand side of the map view. Typically it will look something like this:

This panel tells you all about the currently selected candidates. At the top, you can see how many candidates are selected.

Below that are a number of rows - which rows are visible will depend on what is selected. Each row has a label on the left, and information on the right.

These rows typically display two types of information: categorical values or numeric ones. Categories are displayed as a series of chips, where each chip indicates how many of the selected candidates fall into that category.

Sometimes category rows are elided, to keep the panel short. For example in the image above Category and Name are elided. Clicking on the triangular arrow by the label will show or hide the different categories.

Numerical values are aggregated up in a sensible way, depending on what is being displayed - some will be summed up, others averaged and so on. Most numerical values will show you some more detail (min, max, etc) if you hover the mouse over the value - this is usually indicated by a little superscript blue letter i.

The network candidates panel is at the bottom of the screen. It is a table, listing all of the candidates whose status is not forbidden.

At the top are column headers; below that each row corresponds to a candidate.

When you have some candidates selected you can edit what the model knows about them by pressing the e key (or by right clicking on them).

This will bring up a box with several sections which let you change information about the selected candidates.

Any building can be made into a supply point, and to model a heat network you'll have to make at least one - THERMOS doesn't know about where any supply locations are unless you tell it. If you want to consider a supply that is not in an existing building, you can use the draw building tool to make a new building.

To make a building into a supply, select it and press the s key (or right click and choose from the menu).

This will show the supply parameters box:

The parameters in this box determine the cost and size of the supply for the network model

Maximum capacity

This is the maximum amount of diversified peak load a supply built here could provide. The model may build a smaller supply if that will suit, but never a larger one.

Fixed cost

If any supply is built, no matter its size, this fixed capital cost will be incurred. See above for information about the treatment of capital costs.

Capacity cost

This is a variable capital cost per unit of diversified peak capacity that is built, reflecting the increased cost of building a bigger plant.

Annual cost

This is an annual operating cost per kW diversified peak capacity.

Supply cost

This is the price of producing a unit of heat at the supply. This is multiplied with the system's output and losses each year to determine the cost of heat that was produced.

If you have a fuel cost A and a nominal efficiency B this should probably be A/B

Emissions factors

These determine the emissions the supply is responsible for, per unit of heat output. Similarly to fuel cost, you will probably need to incorporate an efficiency into these values.

Sometimes you may want the model to connect a set of buildings all together or not at all - this could represent a development that has several buildings in it, for example. It can also help simplify the optimisation problem, as the optimiser has fewer decisions to make.

Each building can belong to a single group of buildings - the optimiser will treat that group as a single decision.

To create a group, select a set of buildings and press Shift+G. The selected buildings will be made into a group.

If any of them were previously in other groups, they will be taken out from these groups.

When you have some buildings selected you can see any other buildings which are in a group with any of them coloured in white. You can select everything in the same group by pressing g.

Sometimes you may want to delete something from the map; for the optimiser this is no different than marking it forbidden, but it can make things easier to look at. You can do this by right-clicking on the item you want to delete, and choosing Delete.

The model has two main objectives, selectable on the objective page in the sidebar:

1. Maximise network NPV

   In this mode, the goal is to choose which demands to connect to the network so as to maximize the NPV for the network operator. This is the sum of the revenues from demands minus the sum of costs for the network.

   The impact of non-network factors (individual systems, insulation, and emissions costs) can be accounted for using the market tariff, which chooses a price to beat the best non-network system.

2. Maximise whole-system NPV

   In this mode, the goal is to choose how to supply heat to the buildings in the problem (or abate demand) at the minimum overall cost. The internal transfer of money between buildings and network operator is not considered, so there are no network revenues and tariffs have no effect.

   Under this you can control whether or not to

   • Offer insulation measures
   • Offer other heating systems

   These choices are not relevant when maximizing network NPV, because they can never improve the network operator's NPV so would never be chosen.

Future costs and benefits are combined using geometric discounting. The accounting period controls let you set the number of years into the future to consider, and how much to discount a cost by for each year from the start of the simulation.

Capital costs have two types of special treatment:

Annualizing

capital costs can be annualized, which means converting them into a fixed repayment loan

Recurrence

capital costs can recur, which means that the equipment they represent needs replacing every so often, so the original capital cost is incurred again after this time.

The capital costs section lets you control for each part of a solution:

• Whether it should be annualized and whether it should recur

• If so, a period and a rate.

  The period is used to determine both the term of the annualizing loan, if enabled, and the replacement period if the capital cost is set to recur.

  The rate is used to determine the repayments on the loan, if annualizing is enabled.

If both annualizing and recurrence are enabled, the recurring payments will themselves be annualized, turning the capital costs into a repeating equal yearly cost.

Future payments due to annualizing or recurrence are discounted in the same way as other future costs.

Emissions costs are applied in both whole-system and network NPV modes to anything within the objective which produces emissions. These are things for which you can set emissions factors.

Emissions limits are hard constraints on the solution - if enabled, the optimisation will not be allowed to produce a result where emissions exceed the annual limit.

Often it will be easier for the optimiser to find a good solution if you express emissions goals using costs rather than hard limits.

If you have several possible supply locations on the map but you do not want the model to use all of them, you can set a limit on the number of supply locations the model can build.

For example, if you set this to one the model will only build a single supply.

There are three controls displayed for computing resources:

1. Distance from optimum

   Also known as MIP gap, this tells the optimiser to stop work if it finds a solution which it can prove is within this percentage of the best possible solution.

2. Effect of parameter fixing

   The optimisation involves making an estimate of some key parameters, solving a question assuming these estimates, and then looking at the result to see how good the estimate turned out to be. Here you can tell the optimisation that if this guessing is fairly close to the mark, it might as well stop

3. Maximum runtime

   This setting says how long you are willing to wait for an answer. If the optimisation reaches this time limit without stopping because of the previous two settings, it will terminate. In this case you may get no solution, or you may get a solution whose distance from optimum is worse than the limit you have entered.

At the top of the tariffs page you can define tariffs.

For these to take effect, you must

1. Have the optimisation in network NPV mode - in whole-system mode the 'system boundary' considers only the fuel input for heat, whether that fuel goes into an individual system or a network supply point. The transaction between the building and the network operator is inside that boundary and doesn't affect the objective.
2. Put some buildings on that tariff using the editor controls.

A tariff has three parts:

• A fixed standing charge, which the building pays to the network every year.
• A variable per-kw capacity charge, which the building pays to the network every year multiplied with the building's peak demand.
• A variable per-kwh unit rate, which the building pays every year multiplied with the building's annual demand.

The market tariff is a special tariff whose price is dynamically determined from the other options available to a building.

When a building is offered the market tariff, the model will decide a price by looking for each individual system marked as available to the building at what unit rate for heat has the same NPV as that system (including considering insulation). The market rate is then selected to be the cheapest of these rates, minus a bit (the stickiness) to account for the nuisance of switching.

Each row in the pipe costs table describes the cost of installing a length of flow and return pipe of a given diameter (Nominal Bore). When the model chooses to install a pipe, it will select the first row from this table whose capacity exceeds the diversified peak capacity that pipe needs to carry.

The columns are:

NB

The nominal bore diameter of flow or return in mm. You can add more diameters (rows) using the Add diameter button below.

Capacity

The power that can be transmitted by a flow/return pair of this nominal bore. If you don't fill this in, the cell will show THERMOS' estimate, which is controlled by the capacity & loss model below.

Losses

The heat losses THERMOS estimates for a flow/return pair per meter. Again, you can enter your own values here, or use the computed values THERMOS estimates.

Pipe cost

The cost of the pipe itself, per metre flow/return pair. This is the minimum cost that the model will pay for pipework.

Civil cost

To the right are multiple civil cost columns. These cover the cost of digging a hole for pipework to go in, for a given diameter, depending on the surface or any other local factors.

You can add more of these with the Add civil costs button below, and you can set which civil costs pertain for paths in the candidate editor. Any path for which civil costs are not selected by hand will take the default civil costs set below the table.

Civil cost names are editable in the table header.

To delete a row click the red cross to its right. To delete a civil cost column, click the red cross by its name.

The capacity and loss model controls affect how the capacity and losses columns in the table above are calculated. There is more detail about this in the technical description of the model.

You can see the effect by changing the parameters and seeing what happens in the pipe cost table above. Note that changing the medium or flow / return temperature does not affect the cost of supply or cost of heat exchangers, so you must consider these effects yourself.

Connection costs are intended for covering the cost of equipment within the building, like heat exchangers.

After defining a connection cost here, you can use the candidate editor to assign it to some buildings.

This will determine an extra capital cost, incurred by the network operator in network NPV mode, or by the system in whole-system mode.

THERMOS' pumping costs model is simple: pumping overheads are a fixed share of the kWh output from the system, which may have a different price and emissions factor (since they are produced electrically).

In a heat network, pumping overheads offset some heat output from the plant, since the waste heat produced is still useful. In a cold network the opposite is true, and pumping overheads are added to the output needed from the plant.

The display options at the top of the solution summary page control how financial information is rendered in the tables below. The options mean:

Total

The total un-discounted amount of money spent over the accounting period. This includes all recurring capital costs for replacement equipment, interest payments on annualized capital costs, and all annual operating costs and revenues.

These values are shown in table headers with (¤) as the unit.

Present value

The discounted version of the total cost; all present and future costs and revenues, discounted and summed.

These values are shown in table headers with (¤_PV) as the unit

Principal

(only shown for capital costs) shows just the principal for a single purchase of the item or items, ignoring repeat payments or interest payments on an annualized cost

These values are shown in table headers with (¤₀) as the unit.

Annual

(only shown for non-capital costs) shows the annual cost or revenue for all costs where this makes sense.

These values are shown in table headers with (¤/yr) as the unit.

Below the display options are a series of subsections and sub-subsections which can be selected by clicking on them:

The cost summary gives the headline information the model is working on:

The two most important figures in the cost summary table are the Network NPV (in the middle on the right) and the Whole system NPV (at the bottom right).

These are the values the model will try and maximize, depending on the choice of objective setting.

The whole system NPV will always be negative, because the model only displays costs - the benefit is in terms of delivered heat, which is assumed to be a non-negotiable good desirable at any price.

The pipework section of the network tab lists all the pipe used in the network, broken down by civil cost category and installed diameter.

For more detailed breakdowns you will want to download the results in a spreadsheet and analyse them outside the application.

The demands section lists all demands connected to the network, and shows the cost and revenue associated with them. If they have user-defined fields, you can group them by these fields with the menu in the top left of the table.

The supplies section describes each supply point in the network; the columns are:

Capacity

This is the size of plant needed to meet the diversified peak load

Output

This is the heat output needed from the plant, including losses. In heat network mode this is net of pumping energy, in cold network mode it will include pumping energy.

Pumping

The power required for pumping (see above for how to set pumping costs).

Capital

The capital cost for the plant (units determined by display options above)

Capacity

The operating cost related to plant capacity

Heat

The operating cost related to production of heat (in the output column)

Pumping

The operating cost from pumping energy

Coincidence

The coincidence factor applied to the peak loads this plant is serving.

If the model has installed any individual systems (see above for individual system definitions), this will list the buildings which got them, and their costs.

If the model has installed any insulation on buildings this section lists the quantity, cost and effect of insulation installed.

This gives a breakdown of the emissions in the solution by cause, in terms of their quantities and their costs. Costs are affected by the emissions cost settings in the objective page.

This page shows some technical information about the optimisation itself. The values displayed are:

Objective value

This is the objective value as seen by the optimiser after correcting the diversity and heat loss parameters, but without rounding up pipe diameters.

Runtime

The time spent in the optimiser

Iterations

The number of mixed-integer programs solved

Iteration range

The range in objective value encountered while iterating to find good parameters.

Gap

The MIP gap; this is an upper bound on how close the optimiser is to finding the best possible solution; if zero, the solution is optimal. However, this number is confounded by the effect of parameter fixing. The value shown is the gap for the solution based on estimated parameters.

Bounds

These are the bounds on the optimum value of the best solution found, again confounded by parameter estimation. The true optimum is probably somewhere in this range. Note that the objective value shown above may lie outside this range because it is the value for a slightly different problem in which the parameters have been re-estimated.

Only information from these sheets in the output spreadsheet will be loaded when you import it by uploading:

• Tariffs
• Connection costs
• Individual systems
• Pipe costs
• Insulation
• Other parameters
• Capital costs
• Supply plant
• Supply profiles
• Supply storage
• Supply parameters

These all contain parameters that control the model, but no information about specific buildings or roads.

When you upload a spreadsheet the application will show a dialog asking what parameters you want to use and two other questions:

• Whether to use new parameters where names match
• Whether to keep old parameters with different names

These settings control how existing relationships between buildings and parameters are affected by an upload.

For example, if you have some paths with a civil cost of "Hard Dig" in your problem and you upload a spreadsheet and select to use the pipe costs data, the possible outcomes are:

• If the uploaded spreadsheet contains a civil cost called "Hard Dig" as well:
  • If you have selected to new parameters where names match
    • The paths that previously used "Hard Dig" will now use the new civil cost called "Hard Dig"
  • If you have selected not to use new parameters where names match
    • Existing assignments to "Hard Dig" will be cleared
• If the spreadsheet does not contain "Hard Dig"
  • If you have selected to keep parameters with different names
    • The existing paths will be unaffected
  • If you have selected not to keep parameters with different names
    • The "Hard Dig" civil cost will be removed entirely from the problem

THERMOS is an optimising model which means it makes some decisions.

The main decision is whether to put something in a heat network.

The constraint status of an item affects whether the optimiser can or must use it in a network.

• If the status is forbidden, the item is ignored
• If the status is optional, the item will be included in a network if that is an improvement on any alternative
• If the status is required, the item must be connected to a network if at all possible

THERMOS uses tariffs when it is in network NPV mode to decide how much revenue the network will get from a customer.

The market tariff is a special tariff THERMOS uses, in which a unit rate for heat is chosen that is a bit better than the next best alternative for the building.

Civil engineering costs are one of the costs associated with installing a pipe.

They are to do with digging a hole and filling it back in and making good, so they are affected by the surface being dug, the location of the dig and so on.

They don't include the cost that is the same anywhere in a city: this is the cost of the pipe and mechanical engineering (welding etc) required to fit the pipe.

Connection costs represent the cost of all the equipment required within a building to connect it to a heat network.

Typically this would be a heat exchanger, and maybe some internal pipework.

THERMOS' network model considers only two types of demand:

• The annual demand in kWh/yr
• The peak demand in kW

This is because, by and large, a network connection's size is determined by peak demand, and its operations are determined by annual demand.

THERMOS includes an invented quantity called the "Tank factor".

This exists to reflect the fact that whilst heat network connections are usually sized for hot water production on demand, other heating systems are not. Heat pumps, for example, would have to be very large to deliver the ~20-30 kW of heat needed for instant hot water.

These systems are instead installed with a hot water tank which acts as storage.

The tank factor is used to adjust the peak demand for these systems, by expressing it as a multiple of the minimum possible peak demand (i.e. the annual demand expressed in kW).

Heat network pipes lose heat into the the ground, because they are warmer than the ground.

In THERMOS you can either enter your own losses per metre, or use the built-in model.

If you use the built-in model you probably want to set the ground temperature. If this is higher, the heat losses will be reduced.

Load diversity and coincidence are two sides of the same coin.

As mentioned above, heat networks are usually sized for hot water.

When there are several buildings connected to a network, it is not often a requirement for them all to have hot water on demand at the same time, because hot water usage is a bit stochastic (moreso than heat usage).

Diversity or coincidence is a factor used to approximate this effect.

For example if two buildings have a peak demand of 30kW, we might only supply them using a shared pipe with capacity to deliver 48kW, which is not 30kW + 30kW. This would be a coincidence of 80% - we only size for 80% of the shared demand.

The objective is the technical term for a number that the model is trying to make as large as possible.

For THERMOS this number is always the present value (or minus the present cost).

In network NPV mode it is the present value to the network operator of their revenues less their costs.

In whole-system mode it is minus the present cost of all the costs of delivering the heat.