Progress 11/25/19 to 09/30/20
Outputs
Target Audience:Livestock producers and industry partners. Livestock producers and industry partners were targeted because they would be the end-users of the proposed monitoring system. They also help identify and prioritize the research needs. Last year, the PI reached out to livestock producers, barn builders, and equipment suppliers in South Dakota through face-to-face conversions and extension workshops. Two collaborative research and extension efforts are currently ongoing. Academia. Researchers in the areas of agricultural engineering, animal science, electrical engineering, and environmental engineering were targeted through joint research efforts and presentations at regional and national conferences. This will facilitate interdisciplinary collaborations in developing a holistic solution to environmental monitoring, and ultimately, improved environmental management of livestock production facilities. Students. Undergraduate students were targeted through the PI's Agricultural Waste Management course of which environmental monitoring and control constituted an integral part. Thirty-eight undergraduate students majored in dairy science and agricultural system technology took the course in spring 2019. Many of them will work in the livestock industry after graduation. Graduate students were also targeted by offering professional training and guidance in conducting research projects. General public. This audience was targeted because of a growing concern among the public regarding animal wellbeing and the impact of the livestock industry on the environment. The IoT-based farm monitoring also coincides with the quick development of smart cities and smart homes that are built on similar technologies. The research findings were disseminated to the general public through publications, presentations, and extension/outreach activities. Changes/Problems:The covid-19 pandemic starting in March 2020 has an adverse impact on the project progress. No lab or fieldwork was conducted from March 2020 through June 2020 because of the lockdown of the university campus. Student recruitment, conference participation, and extension/outreach activities were also adversely impacted. What opportunities for training and professional development has the project provided?The project provided great opportunities for undergraduate and graduate students from diverse backgrounds to exchange research ideas, work as a team, and learn Internet-of-Things (IoT), and sensor technologies. Through the project, three students were hired and mentored. Because IoT and sensors are new to most students, the PI has spent on average 6 hours per week working closely to mentor the students. The students' work has resulted in two conference presentations and three prototype sensor nodes or monitoring systems. Two graduate students and one undergraduate student were involved or trained in this project. They are listed as below: Zhisheng Cen is a current Ph.D. student in ABE (Aug 2020-Present). With a background in applied chemistry and agricultural and biosystems engineering, he worked on the development of environmental monitoring sensor nodes and low-cost ammonia sensors. He assisted the PI in setting up and testing IoT networks at the university research farms. Augustina Osabutey is a current Ph.D. student in ABME (Jan 2020-Present). With a background in mechanical engineering and environmental engineering, she helped the PI compare the performance of low-cost particulate and gas sensors in livestock barns. Harsh Dubey was an undergraduate student in electrical engineering and computer science. He helped the PI design and fabricate a LoRaWAN-based monitoring system, including both hardware and software. Harsh started in Jan 2020 and graduated in May 2020. How have the results been disseminated to communities of interest?Three conference presentations were achieved in the first year of the project. Presentations and training were also given through the South Dakota State University extension program to promote the awareness of IoT-based environmental monitoring and management among the public, in particular livestock producers and farm equipment suppliers. What do you plan to do during the next reporting period to accomplish the goals?We will continue to develop cost-effective environmental monitoring solutions for swine and dairy operations. We will keep a close dialog with animal scientists, extension specialists, and livestock producers to identify research needs and opportunities for swine and dairy farm applications. We will also explore the possibility of implementing similar technologies or systems to improve management of livestock, crops, and natural resources. Goal 1: Assess the performance of low-cost particulate matter (PM) and gas sensors through lab and field experiments. We will conduct a four-month field test of the selected PM sensors at the SDSU swine and dairy farms and identify the best performing sensor(s). We will try to modify the PM sensors to improve their performance for coarse particulates. We will acquire low-cost gas (H2S, VOCs, and CO2) sensors and develop a sensor testing platform for field and lab assessment. We will conduct lab testing of the acquired gas sensors and compare their performance against reference analyzers. Goal 2: Develop a low-cost, photoacoustic ammonia sensor. We will finish an initial design, acquire essential components, and build a prototype ammonia sensor. We will start lab challenging experiments to test the sensor output signals under different NH3 concentrations, with and without the presence of other gases. Through the experiments, we will finalize the key design parameters and features of the ammonia sensor and draft a detailed design document. Goal 3: Develop an Internet-of-things (IoT) based monitoring system. We will install and set up two outdoor LoRaWAN gateways on the SDSU research farm to facilitate the implementation and testing of IoT-based monitoring or control systems. We will conduct long-term field tests of the prototype monitoring systems to investigate their reliability, durability, and power consumption. We will develop one or two IoT-based systems for livestock farm management. One system already in our plan is for remote monitoring of light and animal activity in gilt houses.
Impacts
What was accomplished under these goals?
Goal 1: Assess the performance of low-cost particulate matter (PM) and gas sensors through lab and field experiments. (25% Accomplished) Our efforts to assess the performance of sensors was focused on lab testing of low-cost PM sensors. Seven different models of commercial PM sensors (<$50 per unit) were acquired: Plantower PMS5003, Plantower PMS7003, Sharp GP2Y1010, DFROBOT Gravity, Shinyei PPD42, Nova SDS011, and Grove HM3301. They were connected to a microcontroller. The microcontroller was programmed to gather PM concentration readings from the sensors and record the readings, along with date and time (from a real-time clock module), into an SD card. For replicate testing, two such systems were built. Two MiniVol PM samplers were used as the reference method. That is, the PM concentration readings derived from the low-cost sensors were compared to the measurement results from the MiniVol samplers.To further study the impact of particulate size on the sensor performance, a Grimm 11-D aerosol spectrometer was acquired through this project and it classified PM between 0.3 and 20 microns into 31 size ranges. The Grimm spectrometer and Minivol samplers were co-located with low-cost PM sensors during lab testing and challenged with the same PM samples. The lab testing results revealed that all the selected PM sensors, except Sharp GP2Y1010, produced reasonably accurate PM concentration readings (R2 > 0.7) when the PM was comprised of mostly fine particulates (with size < 2.5 microns). However, when coarse particulates became predominant, all those sensors underestimated the PM concentrations by ~23-75% when 50% of particulates by mass were larger than 10 microns, depending on the sensor. This poor performance included the Plantower PMS5003 - the sensor used in Purple Air, the largest non-governmental PM monitoring network in the U.S. Among those sensors, Nova SDS011 delivered the (relatively) best performance - while the sensor underestimated particulate concentration by on average 31%, its concentration data correlated relatively well (R2=0.917) with those derived from reference monitors, thereby allowing for a correlation of the underestimates. A further examination revealed that Nova SDS011 employed a unique airflow design and a relatively high sampling flow rate, which helped reduce the impaction loss of coarse particulates during measurement. Because PM in livestock housing environments (and many other agricultural settings) is comprised of mostly coarse particulates, our findings suggest that many of the low-cost PM sensors on the market cannot be directly used for PM monitoring at livestock facilities. A modification of the sensor airflow channel and sampling flow rate may help improve their performance for coarse particles. Goal 2: Develop a low-cost, photoacoustic ammonia sensor. (5% Accomplished) In the first year of the project our efforts focused on literature review and preparation work. A SolidWorks 3-D model of photoacoustic chambers, a key component of the ammonia sensor, was created. A potentiostat was acquired for high precision, high-frequency current measurement required for sensor development. A phase-lock amplifier was acquired to amplify micro-amp output signals from the sensor's sensing elements. Two commercial ammonia analyzers (Thermo Scientific 17i and INNOVA 1314) were troubleshot and calibrated. They will be used as reference methods for assessing the field performance of the proposed ammonia sensor. The proposed sensor will provide a cost-effective option for ammonia monitoring and control at livestock facilities, thereby promoting environmental stewardship and livestock welfare. Goal 3: Develop an Internet-of-things (IoT) based monitoring system. (35% Accomplished) Our work covered three essential elements of a LoRaWAN IoT network: sensor node, gateway, and webserver. Regarding the sensor node, three prototype air monitoring sensor nodes were built. Each sensor node consisted of a temperature/humidity sensor, a low-cost PM sensor, a semiconductor odor sensor, a microcontroller, a Dragino LoRa radio module, an SD reader, a GPS module, and a power supply circuit. The microcontroller gathered time, location, and air quality data from the sensors, saved the data into an SD card, and transferred them to the Internet through the LoRa radio. We also built one prototype sensor node for remotely monitoring the operating status of the ventilation system of livestock barns, two sensor nodes for remotely monitoring the air temperature in livestock barns, and two sensor nodes for remotely monitoring the water level and temperature in rangeland water tanks. The last two types of sensor nodes employed a low power design. They were projected to be able to run for >2 years without battery replacement, thus reducing the caretaking by farmers. Regarding the gateway, we acquired three commercial LoRaWAN gateways and finished the field installation and setup of two gateway. Each gateway was capable of handling >100 sensor nodes. Because commercial gateways provide little flexibility, we built two self-designed gateways for special application scenarios (e.g. those requiring high data uploading rates) using LoRa radio hats and Raspberry Pi 3B+ microcomputers. Regarding the webserver, we tried three options. The first was a simple, self-developed webserver, including both backend (databases) and frontend (webpages). This option provided the greatest flexibility and allowed us to host the webservers on the university network. The second approach was ThingSpeak, a free, open-source IoT service provider. The third approach was The Things Network (TTN) in conjunction with Cayenne LPP, a free LoRaWAN IoT webserver solution. By integrating the aforementioned sensor nodes, gateways, and webservers, three prototype IoT-based monitoring systems were developed: (1) a livestock barn air quality monitoring system; (2) a rangeland water tank remote monitoring system; and (3) a battery-powered in-barn air temperature monitoring system. The first system has been tested in both lab and field conditions (at the SDSU Swine Education and Research Facility [SERF]). The other two systems have been tested in the lab. All three systems met the performance expectations regarding measurement accuracy, data transmission stability, power consumption, and durability in challenging environments. However, additional tests (e.g. extended field testing) are still needed. Those systems are subjected to further revisions. For example, the sensors selected and/or developed from Goals 1 and 2 will replace the current sensors in the prototype air quality monitoring system. However, through the successful development of the prototype systems, several key design features have been finalized, including systems architecture, communication protocols, and cybersecurity measures. This enables a solid foundation for not only the further refining of the existing systems but also the development of similar LoRaWAN-based livestock management systems.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Osabutey, A., and Yang, X. 2020. Assessment of low-cost particulate matter (PM) sensors for their applicability in the livestock environment. ASABE 2020 Annual Conference. (virtual). Jul 13-15.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Dubey, H., and Yang, X. 2020. Non-disruptive retrofitting of traditional controllers to enable remote monitoring of livestock farm operations. ASABE 2020 Annual Conference. (virtual). Jul 13-15.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Yang, X. 2020. IoT and its applications to CAFOs. Minnkota Annual Meeting. Sioux Falls, SD. Mar 12-13.
|