Introduction: In 2016, the Madison Water Utility (MWU) used 20.5 GWh of energy to pump
nearly 10 billion gallons (PSC 2016). Over the past 15 years, MWU has targeted conservation
efforts and energy efficiency measures that have resulted in an annual average decrease of 1-2%
in annual water production and energy use. Recent graduate student projects funded by MWU have
continued to focus on finding opportunities for further energy savings (Baniel 2013, Hayes 2015).
As a groundwater utility, as much as 99% of total energy consumption can be attributed to pumping
water (Bohnert 2012, Elliott et al 2003), so much of the work has focused on reducing pumping
energy use. This study evaluated the energy savings of variable frequency drive (VFD) installation
on deep well pumps. The objectives of this work were to:
1. Develop a validated procedure to estimate energy savings from VFD installation on a deep
2. Use the developed procedure to identify the strongest deep well candidates for VFD
3. Install and verify the expected energy savings of a VFD on the top-ranked deep well pump.
Background: Over 90% of water utilities in Wisconsin rely on groundwater supplies (PSC
2015b). Energy use by utilities with groundwater supplies is generally higher than that of utilities
purchasing water or using surface water supplies like Lake Michigan due to the elevation that the
water must be lifted. For typical groundwater systems, 99% of energy use can be attributed to
pumping, except for cases where ion exchange or additional treatment is required (Bohnert 2012,
Elliott et al 2003).
Pumping energy use is a function of flow rate, head, efficiency, and run time. For
groundwater utilities like MWU, pumping head or lift is controlled by the drawdown in the well.
To reduce pumping energy use there are four potential alternatives: a reduction in flow rate, a
reduction in head, an increase in efficiency at the pump operating point, or a decrease in pump run time. VFD installation for deep well pumps was investigated to reduce pumping energy use
because reduction in pump speed with a VFD allows utilities the opportunity to control the lift by simultaneously reducing flow rate and increasing pump run time to reduce the amount of
Methods: This work was broken into three separate chapters, with each chapter addressing a single objective. Combined, these chapters provide a comprehensive overview of energy savings from VFD installation on deep well pumps.
The first study focused on the development and validation of a method for characterizing
pump operation and energy use from VFD installation. The developed method used existing pump
operational data from the MWU SCADA system to estimate the static water level, pumping water
level, drawdown, and specific capacity. This data was used in conjunction with friction losses
estimates developed from construction drawings to generate a system head curve. Manufacturer
pump curves and Affinity Laws were used to develop variable speed pump curves. Average
operating flow rate, head, and efficiency were determined for each selected pump speed to estimate energy intensity. Verification of this estimation method was done through field tests at two MWU wells with VFDs currently installed. Tests were run on both well pumps, with the pump run for multiple hours at a range of pump speeds. Pump operational data and power consumption data were collected for each tested pump speed. Observed pump operating points and energy intensities were compared against estimated values to validate the estimation method.
The second study applied the developed method to characterize energy savings potential
across all 22 MWU deep well pumps. Data for a week of pump operation, predominantly from
August 2015, was used to estimate pump operation and energy use with VFD installation at each
site. Energy and cost savings estimates were developed based on MWU’s operational strategy from
2011 – 2015, average daily well production and number of days per year operational. Daily energy
savings were calculated based on deep well pump operation at the most energy-efficient speed
capable of meeting average production. Cost savings were estimated based on a utility-wide
average electricity cost and sites were ranked based on yearly cost savings from VFD installation.
The final study involved the installation of a VFD at the top-ranked deep well pump to
validate cost and energy savings estimates. Pump operation and energy use were characterized
prior to VFD installation. Energy and cost savings estimates for the new pump operating point
after VFD installation were determined using methods from the first two studies. SCADA
operational data and electric utility billing data were used to quantify energy and cost savings from VFD installation. Energy use at surrounding sites was also monitored to determine the impact on energy use elsewhere in the system.
Results: The method to characterize pump operation and energy use for VFD installation was
verified using data from tests at Unit Wells 15 and 25. The developed method was found to capture pump behavior well for variable speed operation. Differences between estimated and observed average operating points were attributed to variations in the system head curve. Changes in static water level were found to be the most significant contributor to changes in the system head curve.
Estimated energy intensities were within 10% of observed values for nearly all tested pump
speeds at Unit Wells 15 and 25. Differences between the system head curves accounted for some
of the difference between estimated and observed values. The differences were most pronounced
at low pump speeds where small changes in head can result in large changes in pump flow rate
and efficiency. For 60% speed at Unit Well 15, observed energy intensity was 14% less than
estimated. A 7-foot difference in static water level shifted the system curve and the observed
operating point had a 45% greater flow rate and 13.5% lower efficiency than the estimated
operating point. Seasonal and long-term variations in the system head curve may make energy use
at low pump speeds hard to accurately characterize. Observed pumping system efficiency was
3.5% – 6.0% less than estimated, but the difference was independent of flow rate.
While energy intensities were generally within 10% of estimates for a given pump speed,
energy intensity savings magnitudes were more variable. At Unit Well 15, observed energy
intensity savings were 50% larger than estimated. At Unit Well 25, savings were 15% greater than
estimated. Variations in the system head curve made estimates of energy intensity savings
magnitudes difficult to predict, but the method was found to be conservative for the combination
of well and pump properties tested.For the second study, pump operation of all 22 MWU deep well
pumps was characterized and curves showing energy intensity versus flow rate were generated for each well, an example of which is shown in Figure 0-1. At Unit Well 30, the minimum
energy intensity was 900 kWh/MG at 70% speed, a savings of 350 kWh/MG compared to 100% speed operation.The most energy-efficient pump speed capable of meeting the average production of 1.8 MGD was 80%. Reduction to 80% speed was estimated to provide a 38% reduction in flow rate, 19%
0 1000 2000 3000
Energy Intensity (kWh/MG)
Figure 0-1 - Unit Well 30 energy intensity curve. Each point corresponds with a 5% speed interval, with 100% speed on the right end of the curve and 65%
speed on the left end.
reduction in head, and an increase of 3.1% in pump efficiency. Energy savings from VFD
installation were estimated to be 500 kWh/day, providing a yearly cost savings of $20,000.
A similar set of results were generated for each of the 22 MWU deep well pumps, and
wells were ranked based on expected yearly energy and cost savings. Under current MWU
operational strategy, the top five candidates for VFD installation were Unit Wells 30, 6, 13, 19,and 11. Estimated yearly energy savings for these sites ranged from 95,000 kWh at Unit Well 11 to 180,000 kWh at Unit Well 30. Cost savings were predicted to be over $10,000 for each of these sites, with a worst-case payback period of 5.6 years at Unit Well 11. A 10-year payback period was determined to be an acceptable return on investment for MWU, and half of the 18 wells
without VFDs at the time of the study met that criteria.
Energy savings potential varied widely throughout the MWU system, and depended on
site-specific combinations of well properties (e.g., static level and specific capacity) and pump properties (e.g., location of best efficiency point). Energy savings potential was also independent of the magnitude of existing energy use - minimal energy savings were estimated for the highest energy use well pumps in the system. Targeting high baseline energy use sites for VFD installation would cause MWU to miss energy savings opportunities at other lower energy use sites. Pump selection played a role in energy savings magnitude. Pumps with an average operating point below the best efficiency point of the pump (greater flow rate and lower head) had increased energy savings magnitudes.
The final study analyzed actual cost and energy savings after a VFD was installed at Unit
Well 30, the top-ranked candidate from the second study. Average site production was unchanged
by VFD installation and speed reduction, 1.9 MGD. Observed benefits from VFD installation and
pump operation at 1,450 gpm, down from nearly 2,400 gpm, were in line with estimates. Deep
well power consumption was reduced by 102 kW for an energy intensity savings of 335 kWh/MG,
both values were within 5% of estimated savings. Energy use at Unit Well 30 was reduced by 540
kWh/day compared to the previous billing period, for a cost savings of nearly $65/day. After six
months of operation (December 2016 – May 2017), MWU has saved 94,400 kWh and $8,650
compared to the same period in 2015-2016. Payback for VFD installation at Unit Well 30 should
be within 1.5 – 2 years.
Energy use and site operation at Unit Well 18 and Booster Station 118 were monitored to
determine if there was increased energy use increased elsewhere in the distribution system after
VFD installation. Average production, energy use, and energy costs were nearly unchanged at Unit
Well 18. The only changes were increased number of days operational and a one hour per day
increase in booster pump run time. Booster Station 118 was sporadically used before and after
VFD installation, and there was no evidence of adverse effects from changes at Unit Well 30.
Changes in operation at Unit Well 30 did not yield increased energy use or costs elsewhere in the distribution system, allowing MWU to benefit from the full $65/day of cost savings at Unit Well 30.
Energy billing information at Unit Wells 18 and 30 highlighted the impact of demand
charges on cost savings for MWU. Electric utility demand charges made up nearly 40% of the bill
at Unit Well 18 and nearly 40% of cost savings at Unit Well 30 were attributed to demand
reduction. Demand reduction was a crucial component of cost savings from VFD installation. An
increase in pump speed for less than a week at Unit Well 30 in April 2017 cost MWU $300 of cost
savings through increased on-peak demand charges. Estimation of cost savings from VFD
installation based on an average electricity cost overlooked the importance of on-peak demand