Kyocera Solar Inc.
7812 East Acoma Drive
Scottsdale, Arizona
Solar powered pumping systems using centrifugal pumps have generally been difficult to design and specify. Established methods of centrifugal pump selection rely on constant motor speed to simplify the pump selection process. Solar systems operate with varying power levels and therefore varying speeds. The optimal choice for a pump becomes a complex function of geographic location, solar resource, system power over time, well depth and desired daily water output. Previously, optimal pump selection involved trial and error simulations by engineers or professionals with access to specialized software.
A system will be demonstrated that simplifies the pump selection process. The method presented is based on measured data and computational models of world climate, pump, motor and solar array performance. The method uses an easy to interpret system of maps and charts and provides a quick, simple and accurate means for designing a solar pumping system and selecting the components. Knowing only the geographic location, depth to water, and desired daily output, the proper pump and solar array can be determined in minutes without the need of a computer. For the first time, resellers, installers, and more importantly, customers have the ability to design and specify a pumping system without the experience of an engineer.
Water is one of the necessities of life. Obtaining that water through traditional means such as fossil fuel generators can present impractical challenges in many areas of the world. Be it for potable water, water for irrigation or livestock, a demand for solar water pumping exists in those areas where utility power is unavailable or distant. Solar power may provide the only economic and reliable means to pump water. In many cases, those who can benefit the most from solar water pumping are also the least able to afford it. The need for efficient, optimally designed systems is paramount.
Solar power varies throughout the course of the day; zero at dawn and dusk, peaking at solar noon. To complicate matters further, the solar modules are affected by temperature, and the power output falls as the solar cell temperature rises. Energy production, and therefore water yield, varies during the course of the year, generally reaching maximum levels in the summer when the days are longest. With the exception of small systems, the solar panels are the primary cost component in a solar water pumping system. In many cases tracking the solar array in one, or sometimes two axes provides an economic means of obtaining more water or reducing the size of the solar array in a given system. The tracker orients the solar array towards the Sun in the morning, and follows the Sun during the day. This allows the pump to operate earlier in the morning, and later in the evening. Over sizing the solar array accomplishes very similar results, and while costly, can be considered in cases when a tracking is not possible.
Patterns of usage also vary. Agricultural applications often benefit from the increased water production in the Summer. In these situations, there is a convenient coincidence between production and demand. Other situations, such as water for drinking and bathing, require that water production remain more constant throughout the seasons, and that a minimum requirement can be guaranteed during the Winter. Many systems are designed to operate at less than full rated pump power to reduce the size of the solar array and system cost. Traditionally, predictions of pump output and daily water production are made using models of temperature, solar radiation, and piecewise integration of pump output. These methods have proven sufficiently accurate. However, selecting the proper pump has been more awkward.
A multistage centrifugal pump, the norm in water well application, operates at peak efficiency over a relatively limited range of pressure or depth. Figure 1 illustrates the dramatic fall in efficiency when the operating depth is either too high or too low. To address this, pump manufacturers typically offer a progression of different pumps serving a range of well depths and output requirements. Deeper wells require more stages in the pump to optimize efficiency. Simply stated, the deeper the well the longer the pump. Multistage pumps that run from AC power generally operate at a nearly constant rotational speed. For a given well depth, this allows simple selection of the proper pump. This simple situation is complicated rapidly when direct solar power is used to drive the pump.
Figure 1: Measured Pump Efficiency vs Depth for six different pumps
As shown in Figure 2, the optimal pumping depth of a given pump is roughly proportional to the applied power. Because of the varying power levels pump selection is typically done by trial and error. Simulations of daily output for a given solar array and location are performed for several pumps. The pump yielding the greatest output is chosen. Experience and educated guesses allow the number of different pumps simulated to be reduced. Simulations with differing solar sizes may be required to insure that the minimum specifications for daily or seasonal output are met. However, few users have access to the facilities or the training to choose the right pump, solar array, and predict the system performance.
Figure 3: 3D representation of measured data for one pump, the measured current and voltage are combined to form power
The second step in the process involved system simulations using commercially available solar system sizing software. The software creates performance simulations based on location, system head, and pump and array selections. The software requires the input of pump and weather models for given location in tabular form. An artificial location with discrete levels of solar radiation was created. These simulations show daily and hourly output for a selected system. The software, does not however, perform any component selection; simulations are performed only for user specified locations, solar arrays, and pumps.
Hundreds of simulations were run. Using the artificial location, all pump models were simulated over a range of heads and array sizes. The results of these simulations were collated and represented in chart form as shown in Figure 4b. The charts were derived by simply determining what pump performed the best under a given set of operating conditions. The results show that proper pump selection is affected by far more than just depth of operation. Solar resource, and array size also play an important role in pump selection.
Sizing a system starts with determining geographic location. From there a color-coded map provides the Sun Hours On Tilt (S.H.O.T), and the optimum tilt angle for the solar array.
Figure 4a: S.H.O.T. in Map form (Maps have been calculated for the entire World)
Figure 4b: Sizing chart for regions of the World with 5-6 S.H.O.T. and requiring Summer water
A simple worksheet is used to determine the effective total dynamic head taking into account pipe resistance. It is then simply a matter of selecting the pump-chart (color-coded to location) and finding the correct pump and array size from the required water and head. The charts are marked in both imperial and metric to make selection possible without the user having to resort to conversion factors.
The Solar Water Pumping Applications Guide published by KSI contains all the charts and examples for the Kyocera pump line, as well as similarly simple charts to size a system based on the SD-series of low-cost diaphragm pumps.
So, in minutes, without the need for a computer, a dealer or end user can correctly size solar water pumping system.
We have developed, validated and distributed a simple method for sizing solar water pumping systems based using either centrifugal or diaphragm pumps. The method has been demonstrated to several dealers during seminars and has been received well. We have started to produce pump curves in terms of power (rather than current) as requested by one of our customers. This entails fitting the 3-Dimensional data presented in Figure 3 to extract Flow = ƒ(Power, Head). This has been achieved for the ½ horse-power motors, and will be extended to all systems. This work has also led to discussions on how to produce sizing charts for fixed voltage (i.e. battery powered) pumping systems. Climate data is normally only available as hourly averages. Using commercial software to estimate the output has so far been based on this rather coarse time interval. Finer time resolution for the integral would provide a better estimate of water output, especially at lower intensities in the morning and evening. A more accurate representation of the Maximum Power Point Tracking (MPPT) controller is also required to reflect the actual operating parameters the KSI designed controller uses. More information is available at the Kyocera Solar web site.
A version of this paper is available on the WWW at: http://www-rcf.usc.edu/~arjones/KSI/pump-poster.html
The first step is to measure whatever can be easily measured. This is ok as far as it goes.
The second step is to disregard that which can't be easily measured or to give it an arbitrary quantitative value. This is misleading.
The third step is to presume that what can't be measured easily really isn't important. This is blindness.
The forth step is to say that what can't be measured really doesn't exist. This is suicide.