Inflow and Infiltration Shows Up in the Pump Runtime

A utility with fourteen lift stations and one treatment plant knew it had an inflow and infiltration problem because the plant flow tripled during storms. What it did not know was where the water was coming in. Flow meters at every station were quoted at a price that would have paid for a good deal of pipe lining, and the meters would have told the utility what it could already have learned from numbers it had been collecting for six years. Every station reported pump runtime and start counts to SCADA. Nobody had done the arithmetic.

Runtime is a flow meter you already own

A lift station pump, when it runs, moves water at a rate set by the pump and the system head. If that rate is known, the runtime for a day multiplied by it is the volume pumped that day, which is the volume that came into the station. The estimate is not as good as a flow meter, because the rate varies with wet well level and with wear, and because two pumps running together deliver less than twice one pump. It is good enough to compare a station with itself across weather, which is all I&I ranking needs. The daily runtime total is in the historian, or in the runtime counters that reset each day, or can be built from the start and stop events with a query.

Get the pump rate from a drawdown test

The nameplate rate is what the pump did on the factory test at a particular head. The rate that matters is what it does in this wet well against this force main today, and a drawdown test measures it. With inflow steady, turn the pumps off, let the well rise between two level marks, and time it: that gives the inflow rate from the well geometry. Then run one pump and time the fall between the same marks: the pump rate is the fall rate plus the inflow rate. Twenty minutes, a stopwatch, and the wet well dimensions. Do it for each pump, do it after any pump or force main work, and store the result with the station record, because a pump rate that has fallen twenty percent from last year is a maintenance finding in its own right.

Build the baseline

For each station, take the daily volumes for a stretch of dry weather, ideally a couple of weeks with no rain in the preceding week, in each season. The average is the station's dry-weather flow, and it is also a sanity check: it should be close to the number of connections times a typical per-connection flow, and a station whose dry-weather flow is far above that has a baseline infiltration problem before any storm. The diurnal shape is useful too. Runtime by hour on a dry day shows the morning and evening peaks of a residential basin; a station whose night flow is nearly as high as its day flow is taking in groundwater.

Read the storm

Line up each station's daily volume against rainfall from the nearest gauge and the patterns separate themselves. The ratio of the peak wet-day volume to the dry-weather baseline is the number to rank by. A station at 1.5 times baseline in a big storm is doing well. A station at five times baseline is where the crew should go. The volume above baseline, summed over the storm, is the gallons the utility is paying to pump and treat that fell out of the sky.

Rank, then look

The ranking is the point of the exercise. With fourteen stations, the runtime analysis at the utility above put three of them at four to six times baseline and the rest under two. Those three basins got the smoke testing and the camera, and two of them turned out to have most of their problem in a few hundred feet of pipe and a dozen roof drains. The flow meter budget went to lining. The method does not find the leak; it tells the utility which square mile to look in, and that is the expensive part of the search.

Setting it up in SCADA

1. 1

   Confirm the runtime tags

   Each pump needs a runtime that accumulates reliably, either a daily total that resets at midnight or a lifetime counter that the report differences. Check that the counter does not reset on a controller power cycle.

2. 2

   Store the pump rates

   A per-pump rate tag, entered from the drawdown test with a date, so the report can multiply. Put the date on the screen so a stale rate is visible.

3. 3

   Build the daily report

   Volume per station per day, with rainfall alongside from a gauge tag or a manual entry. A spreadsheet export works if the historian cannot do the arithmetic.

4. 4

   Compute the ratio

   For each station, volume divided by the dry-weather baseline. Trend it. The stations that jump are the list.

5. 5

   Keep the baselines current

   Recompute dry-weather flow each season, and after any change in connections, so growth is not mistaken for infiltration.

Where the method breaks

The estimate is only as good as the pump rate, and pump rates drift. A station whose runtime creeps up over a year with no change in rainfall may have a clogged impeller rather than a new leak, and the drawdown test tells the two apart. Two pumps running together in a storm deliver less than the sum of their rates, so peak-day volumes are undercounted at exactly the stations that matter most; a second drawdown test with both pumps running fixes that. And a station that surcharges or overflows in a storm has runtime that is capped by its capacity, so its ratio understates the problem. Those are the stations where a real flow meter, or a level-based overflow estimate, earns its cost.