Skip to content

Worked example: Bello

Bello sits in the Medellín metro area of Colombia, and its sanitary network has a problem money can see: the treatment plant is across the Medellín river and uphill, so every drop of wastewater has to be pumped. When lifting is the whole job, the pumps are the dominant operating cost - and unlike a pipe, that cost recurs every single day.

Where Warkworth asks is the model right, Bello asks is the operation good. Same hydraulics, a different question: given what the pumps are doing now, what would it take to move the same volume for less money?

Network~2,390 junctions and conduits - a genuine utility-scale sanitary system
Forcingdry-weather sanitary flow ([DWF]) at every node, on a daily pattern
Pumpingtwo pumps, P1 and P2, lifting flow toward the plant
Dischargeone outfall, OF1 - the treatment plant across the river
UnitsSI (LPS / ha / mm); real Medellín geography, opens on the Streets basemap

Run the model (the green Simulate button; give it up to a minute or two - it is a big network) and open the Pumping Summary in the run report. That is where the operating cost lives:

PumpUtilisationAvg / peak flowEnergy over the day
P199.99%82 / 127 LPS157.2 kW-hr
P299.93%9 / 14 LPS32.6 kW-hr
Total~7.5 ML lifted~190 kW-hr

Two things to read off that. First, the pumps are utilised ~100% of the time - they run essentially flat out, all day, because there is no storage to let them rest. Second, that is ~190 kW-hr to move 7.5 million litres to the plant. Multiply by 365 and the annual energy is the number a utility’s finance team actually feels.

SWMM computes energy (kW-hr), but the bill depends on when you use it. Colombian water utilities buy power under CREG’s Tarifa Reducida: they pay 80% of the full industrial rate and are exempt from the 20% contribution surcharge that other industrial users pay. On an illustrative time-of-day tariff - full rate $0.22/kWh peak, $0.14 off-peak - that makes the utility’s effective rate $0.176 peak / $0.112 off-peak. Bucketing the baseline energy by clock time:

Baseline - pumps run continuously ($8,857/year):

WindowEnergy/dayAnnual energyRateAnnual cost
Peak (18:00-22:00)46.3 kW-hr16,900 kW-hr$0.176$2,974
Off-peak143.9 kW-hr52,500 kW-hr$0.112$5,883
Total190.2 kW-hr69,400 kW-hr$8,857

About a quarter of the energy lands in the four-hour evening peak - the pumps run through it because flow arrives and must leave. That is the money left on the table.

belo-optimized.inp implements the fix. It converts each pump’s inlet to a wet-well storage node (sized to hold the four-hour evening peak - about 1,700 m³ at the main pump) and adds [CONTROLS] that hold the pumps during 18:00-22:00 while the wells fill, then run them off-peak to draw the wells back down. A start/stop deadband (2.0 m on, 0.3 m off) keeps the pumps from chattering, and an emergency rule runs a pump if its well ever nears the top. Same network, same ~7.5 ML/day to the same plant - only when it is pumped changes. It runs clean (continuity 0.16%).

Optimized - pumping shifted off-peak ($6,435/year):

WindowEnergy/dayAnnual energyRateAnnual cost
Peak (18:00-22:00)0 kW-hr0$0.176$0
Off-peak157.4 kW-hr57,450 kW-hr$0.112$6,435
Total157.4 kW-hr57,450 kW-hr$6,435

Saving: about $2,420 a year (27%), on ~17% less energy - achieved without moving a single pipe.

  • The tariff numbers are illustrative. The $0.22/$0.14 peak/off-peak split and the 18:00-22:00 peak window are stand-ins; the CREG 80% reduction and surcharge exemption are real, but plug in your own rates and hours to price it for a specific utility.
  • Part of the saving depends on the wet-well design. The timing saving (~$1,080/year) is robust; the lift saving on top of it scales with how deep the wells are, so the ~$2,420 headline is design-dependent, not a fixed law.
  • The energy saving alone does not justify the storage. Building ~1,700 m³ of wet well is a large capital cost, and $2,420/year would take a long time to pay it back. Storage like this is normally built for equalisation and reliability, with the energy saving as a co-benefit - this example quantifies that co-benefit, it does not claim the storage pays for itself.
  • Sanitary dry-weather flow only - no wet-weather inflow or rainfall-derived infiltration. This is about steady pumping economics, not storm response.
  • Representative geography. The network is derived from open map data over the real Medellín terrain, not an as-built utility record - a realistic system to reason about, not a survey.