Partners in Health (PIH) provides quality healthcare to some of the poorest people in the world. With the original focus to serve Haiti’s mountainous Central Plateau region, PIH has established 11 healthcare facilities that serve more than a half million people—regardless of their ability to pay. Only one of these facilities is tied to an electrical grid, while the others rely on diesel generators and inverter /charger battery systems to power the hospitals’ loads. The cost of diesel fuel—both the purchase price and the delivery price—is one of the largest budget items, taking funds away from healthcare needs, such as medicine and clinical personnel.
In September 2009, the Solar Electric Light Fund (SELF) partnered with SunEnergy Power International (SunEPI) to provide solar electricity to the first of the PIH Haitian facilities in Boucan Carré, located in the central plateau region. As part of the installation, USAID donated funds for a technician training led by Walt Ratterman, Herb Kanski, Carol Weis, and Christopher Freitas. Twenty-one hospital technicians from around the country attended the training, including a Haitian technical college professor looking to bring renewables into his program in Port au Prince. In addition to the nine-day technician training, there was a half-day training given to the medical staff to educate them on the abilities and limitations of the PV system.
Even in Boucan Carré, a town of nearly 50,000, there is no utility power. Most homes have no electricity, and only a few benefit from generator power.
The local hospital—run by Zamni Lasante (the Creole name for Partners in Health) — had been powered by a 35 kW diesel generator, along with an aging, undersized central inverter/charger battery system comprised of two Trace DR modified square-wave inverters with four paralleled strings of Trojan T105 batteries.
To serve the hospital and dormitory loads, including lights, fans, lab equipment, computers, and the water pump, the generator typically ran from 9 a.m. until 5 p.m., and then started again at night to service the dormitory loads for several hours, for a total daily run time of 10 to 12 hours. The old DR system powered the staff dormitory and hospital ward loads for several hours each night until the decrepit batteries discharged too much to operate the inverters. Fuel for the generator is hauled to the hospital in 55-gallon plastic drums in the back of small, four-wheel-drive pickup trucks, as the river and the rutted road does not allow a tanker truck delivery.
The new PV system needed to achieve three specific goals:
To achieve the first goal, a true sine wave inverter was installed with appropriately sized batteries to carry a significant portion of the hospital’s loads for most of the day. The old inverters were decommissioned from this site to be used at another clinic, and the well-used batteries were recycled.
An additional step of isolating sensitive laboratory loads from the generator was necessary. During the middle of the day, and under normal operation, the generator was needed to augment solar-charging to the batteries and run some of the larger loads. When the generator runs, all loads—including sensitive ones—connected to the inverter/charger system are automatically transferred to the generator. Without regular maintenance, generator electricity can easily fall out of voltage and frequency specifications. Plus, it is typically not considered “clean” enough for the hospital’s sensitive lab equipment, resulting in premature failures or malfunction of equipment.
To operate some of the costly lab equipment used to diagnose HIV/AIDS (one of the leading causes of death in Haiti), a stable, clean sine wave, along with tight voltage and frequency windows, is needed. To serve this need, one inverter was designated as a “no-generator-contact” inverter, meaning that the inverter and the loads it powers would never be connected to the generator and therefore be ensured of providing cleaner electricity.
In addition to the no-contact inverter loads, there were also “generator-contact” loads at the hospital. These loads would be connected to the PV/battery system through a separate inverter, but when the generator was on, they would be allowed to connect to it without worry of damage.
Finally, some loads—ones that had large draws and would cause excessive battery cycling—were isolated on a generator-only panel.
Achieving the second goal—saving money by cutting fuel costs—was not straightforward. Since the generator was still needed to augment battery charging and to run larger loads, the challenge was figuring out when it should run.
Generator specifications show that the relationship between fuel consumption and kilowatt-hour load is not linear. True cost savings occur by reducing overall generator run time, as well as creating conditions so the generator always runs at 50% capacity or more. The important analysis to be done was to determine if all large generator-only loads could be coordinated. Further design details focused on accurately programming the controls to ensure that, when the sun was shining, the PV system would take precedence in charging the batteries. At the project’s onset, the optimistic goal was to reduce generator run time by 50%.
And last, addressing the critical nature of services the clinic provided to the remote area was important. With limited access to the clinic in severe weather—plus the long history of devastating hurricanes and political unrest—having an independent, on-site energy source would allow the hospital to continue providing medical services during the most desperate situations.
The hospital’s original rough load analysis, done a year before the system’s installation, estimated that the new inverter/charger system would need to handle at least 47 kWh per day. This figure anticipated an additional amount of unknown dormitory and auxiliary loads, all of which were typically connected to the generator for 10 to 12 hours each day.
This rough estimate was used to ballpark a PV system size, with the understanding that donated equipment was only going to offset part of the hospital’s loads. On the second day of class, the students interviewed staff members about the number of hours the loads were typically used and traced every load on each circuit breaker. The resulting energy analysis was more than double the original estimate and provided enough information to determine which circuits should stay on the generator-only service panel, and which ones could be moved to the new PV system, as either contact or no-contact loads.
The system includes 66, 155 W SolarWorld PV modules mounted on a locally made custom rack on the hospital’s main roof. To deter thieves, special hardware covered by welding steel plates over them locks the modules in place, and the rack base was welded to rebar and attached to the roof with concrete.
The PV modules were split into four subarrays, each with its own combiner box that routed to four separate MPPT charge controllers. The modules were wired in series in groups of three to match the controller input voltage window. Each positive conductor from the series string was wired to a 15 A circuit breaker in the combiner box. In the combiner box, all the series strings connected in parallel and a larger wire, carrying all the ampacity, continued in conduit from the roof to the charge controller.
All of the balance-of-system equipment (inverters, controllers, and batteries) was located on the west side of the hospital’s equipment room. The batteries were located directly behind the inverters in another separate room to prevent battery gases from corroding the electronics. The 24 2-volt, 1,766 Ah batteries were wired in series, making an 84 kWh battery bank.
By the fifth day, the PV system was installed and charging the batteries, and the group focused on setting up an on-site monitoring system. Since this system served all the hospital workers, the staff dormitories, and the immediate local residents, all participants needed to understand system monitoring and load limitations. It was decided that having signal lights in a highly visible area was important to show the battery’s state of charge (SOC). This allows the larger community to determine if they could continue using loads, if the generator needed to be run, or if loads needed to be reduced. A green and a red light were used to indicate battery SOC, with green signaling 90% and red signaling 50%. The typical protocol was to run the generator from 10 a.m. to 3 p.m., provided there were no emergency medical loads. On a sunny day, the PV system would start charging the batteries at about 8 a.m. By 10 a.m., the generator would be turned on, which would carry most of the hospital loads and any large loads, like air-conditioning and water pumping. This would protect the batteries from deep discharge, and the programming of the PV charge controller and the inverter/charger allowed the PV array to continue charging the batteries even with the generator running. At 90% SOC, the green light would turn on. If no emergency surgical loads were needed, one of the technicians could shut down the generator, with the PV array continuing to charge the batteries until they were full.
Although the deep-cycle lead-acid batteries used in this system can be discharged to 20% SOC, the signal light and audible alert was programmed to come on at 50% SOC, to promote conservation of all non-critical loads at this juncture, to allow some reserve power to remain in case of an emergency, and to allow longer battery life. Under normal conditions, the technician disengages the alarm and starts the generator to charge the batteries. If the generator is not functioning, the procedure is to shut off all non-critical loads and wait for the PV system to charge the batteries.
After allowing the batteries to charge with no loads connected for two days, the hospital loads were activated at noon on the third day. By 4 p.m., with thunderclouds accumulating, battery SOC registered 99%. At 8:30 the following morning, the monitoring system showed the battery SOC was 73%.The solar gain later that morning had upped it to 76%.
In theory, if all nights were similar to the test run, and the technician assured that the battery reached at least 90% SOC each afternoon, there would be sufficient battery capacity until the next day.
The second night showed significantly more loading. By 4 p.m., SOC had dropped to 91%. The next morning at 8:30 a.m., contrary to all of our best guesses, the battery minimum SOC showed a low of 47%, with the current SOC at 49%.
The second night’s increased demand, which used 44% of the battery capacity, prompted further load investigation. More wires were discovered that were traced to outside buildings not originally part of the load analysis. Two inefficient refrigerators in an exterior kitchen, as well as several circuits running lights and outlets, were identified, and alternatives were discussed with the group.
The daily operation duties chart and logging sheet prompts system technicians to record the time, weather, battery SOC, battery volts, the kWh meter reading, whether the alarm and red light activated, and the number of hours the generator was run that day. It also provides a simple decision matrix of when to turn the generator on and off according to varying conditions.
Normal procedures dictate that the technician checks the battery SOC at 7 a.m. If the weather is sunny and the SOC is above 60%, the technician waits until 10 a.m. to start the generator. If it is cloudy, the technician starts the generator to sustain morning hospital loads. At 3 p.m., the technician verifies whether the green light is on, indicating the battery is at 90% SOC and, if so, shuts off the generator. In case of signal light failure, the technician also checks the SOC on the Mate to verify battery SOC has reached 90% before turning off the generator.
A monthly maintenance agreement was originally drawn up with a local RE company, Green Energy Solutions, to check battery water levels, inspect the system, review daily logs, and do brief educational trainings with the local technicians, as necessary. SELF has since hired a technician to do monthly service work at this and other hospitals the organization is electrifying.
In mid-November 2009, after two months of the system’s operation, Walt Ratterman returned to Boucan Carré to compare the system performance with project’s original goals and learned that:
The hospital staff has learned a lot about system management, and is making progress in learning how to keep accurate logs on the system’s operation.
The old, inefficient refrigerator and freezer had been replaced with two efficient refrigerators. A vaccine fridge that was previously run on liquid propane gas was also replaced, saving more fuel. The staff had removed some of the loads going to non-essential, off-site locations.
The ability to monitor performance data on a regular basis is an important element in keeping the energy systems functioning properly, and in catching problems before they become costly to correct. The system monitor currently records operation but the data can only be accessed on site. SunEPI and SELF are now working together to implement a remote monitoring system for Boucan Carré. Once completed, this monitoring system will allow the system’s data to be accessed over the Internet.
Among all the tragedies that occurred during the January 2010 earthquake, both Walt Ratterman and Herb Kanski died in Port au Prince while meeting to design the next nine public hospitals. Both their spirit and their huge knowledge base has been one of the many devastating losses of this event. Fortunately, the Boucan Carré system did not sustain any damage, and is currently providing electricity to doctors serving hundreds of patients a day. The lessons that are learned from this system continually guide the designs and protocols for the next rounds of hospital PV systems.
Carol Weis continues her work as a renewable energy educator in Haiti and other countries, bringing sustainable energy to rural health clinics and schools. Carol works part-time for Solar Energy International and is a project engineer for SunEPI. She is a certified ISPQ Master PV Trainer and holds a NABCEP PV Installer Certification. She has worked as a licensed electrician and solar installer in Colorado.
Walt Ratterman, SunEPI’s CEO, was tragically lost to the Haitian earthquake on January 12, 2010. Walt’s hands-on PV experience included residential and commercial PV installations in the eastern United States, as well as rural PV installations in Nicaragua, the Galapagos Islands, southern Ecuador, Peru, Arunachal Pradesh in India, Burma, Thailand, Haiti, Rwanda, and many other countries. Much of his work is available as a resource on both SunEPI and Powering Health websites (see below) His guidance, humor, and dedication to helping the less fortunate is missed dearly throughout the world.
Donations: Modules and power equipment were donated by SolarWorld, and training and local installation costs were underwritten by USAID. The Good Energies Foundation provided additional support. SELF arranged for the purchase of batteries and other equipment. Walt Ratterman of SunEPI and Herb Kanski of Tetra Tech led the training and installation team in Haiti, and managed all of the complicated logistics, local installation, and translation efforts.
Partners In Health • www.pih.org
Powering Health • www.poweringhealth.org
Solar Electric Light Fund • www.self.org
Solar Energy Power International • www.sunepi.org
OutBack Power Systems • www.outbackpower.com • Inverters & monitoring
SolarWorld • www.solarworld-usa.com • PV modules
Surrette • www.surrette.com • Batteries