Evaluating Environmental Monitoring Applications of Low-Cost Sensors for Electric Utilities A look at the potential applications for new low-cost sensors in industrial environmental monitoring.
A variety of novel environmental monitoring technologies, such as small low-cost air quality sensors, have been on the rise for the past several years, fueled in part by advances in manufacturing technologies that have eased miniaturization of electronics. These devices are often quite inexpensive and accessible to potential users, while still providing real-time information on environmental metrics. Sensor tools have the potential to provide screening-level data in currently unmonitored areas, or to supplement more complex existing monitoring programs. However, their applicability for environmental monitoring performed by electrical utilities has yet to be vetted. This article considers potential applications of sensors to a subset of industrial facility monitoring that is occurring at electric utility sites. Potential for Sensors in Industrial Environmental Monitoring Scientists and engineers at our organization, the Electric Power Research Institute (EPRI), are always looking for new tools and processes that may help electric utilities operate their facilities more efficiently, with more flexibility, and at lower cost. Over the past several years, EPRI has engaged in a crossdisciplinary research program on intelligent sensor systems and associated data analytics that is investigating sensor performance and data acquisition, data communications and manipulations, and final use of sensor data in facility operations (see Figure 1). Sensor applications are considered with relevance to the various electricity research sectors: generation, nuclear power, transmission and distribution, and energy and environment. Intelligent sensors are needed throughout the power system to transform raw data into actionable information. For example, EPRI has developed its own sensor packages for equipment monitoring and condition-based maintenance. These devices are being tested on the transmission grid and at substations to demonstrate the technology s potential to reduce or extend intervals between preventive maintenance and surveillance tasks. Development and in-plant testing has been conducted of novel systems that can continually sense torsional vibration at turbine shafts. This enables early detection of conditions that cause turbine blades and other rotor elements to fail. Similarly, chemical or physical sensors might be relevant tools for environmental monitoring at power system facilities. In addition to their relatively low cost, the ability to deploy sensor systems in complex environments and without line power make them attractive for locations that cannot accommodate traditional instruments due to space or infrastructure limitations, such as on the grounds of working industrial facilities. Ideally, real-time information about electric power system assets provided by sensors would be secure, and incorporated to operate the system efficiently and effectively while managing environmental impacts. EPRI scientists have brainstormed a variety of potential Figure 1. EPRI s approach to electric utility sensor systems.
environmental applications relevant to electric utilities for which they would like to see viable sensor options exist. Some examples include: 1. Use of sensors to help site permanent monitoring instruments; 2. Measurement of emissions sources at power generation facilities (e.g., fugitive dust or methane); 3. Creation of early warning or detection systems, such as for geotechnical parameters (e.g., berm stability) or leaks of materials in confined spaces; 4. Worker personal exposure monitoring; and 5. Interaction with local communities or other stakeholders for education. Listing of these example applications does not necessarily mean that the technologies are currently commercially available and appropriate for deployment now, rather they are aspirational applications if sensors of appropriate performance are identified. Real-World Testing Is Vital for Sensor Evaluation As with any new and emerging technology, it is crucial to ensure that the performance capabilities and limitations of sensors are adequately understood before they are applied. These complexities can then be communicated to stakeholders. Despite a marked increase in recent reports summarizing results from field studies testing environmental sensors, a lack of evaluation data still exists for many potential applications. Additionally, the fundamental designs of low-cost sensors make it difficult to extrapolate performance from one site, metric or sensor model to other situations, even if they are generally similar in design. Prior research studies have repeatedly shown that while sensor components themselves can often have high precision, many of the detection techniques used can be subject to chemical interferences (cross-sensitivities), drift, and other factors that affect data quality to a greater degree than traditional instrumentation. Even if these concerns may be anticipated due to manufacturers specifications and prior laboratory analysis, they do not always present in expected ways due to the complex real-world ambient air matrix in which they are deployed. Therefore, comprehensive sitespecific performance testing can be used to determine suitability of sensors for electric utility applications. A further need in the design of sensor performance testing is a robust assessment of the full sensor system cost. For example, the capital cost for the sensor component and the package into which the sensor is incorporated, including electronic control boards, power systems, wireless communication systems, and data handling infrastructure, all need careful evaluation. Additionally, operations and maintenance costs for sensor deployments are often expected to be relatively low, due to the relative ease-of-use of sensors compared to traditional reference instrumentation. However, the labor cost of any need for frequent maintenance (e.g., online or offline calibrations or cleaning activities) should be a consideration. Management of the challenging big data files resulting from the real-time measurements can also be substantial. Any evaluation of whether low-cost sensor technologies are an appropriate opportunity to pursue for a given application should, at minimum, consider these issues that contribute to their classification as a help or a hindrance to facility monitoring. Lessons Learned from Test Applications Several EPRI pilot projects have begun to test the capabilities of individual environmental sensors, as well as logistics concerning their usage, such as power supply (often off the grid) and wireless data transmission. One pilot project involved a deployment of particulate matter sensors for fugitive dust measurement at a power generation facility. Dust sources at these facilities can include material handling processes for coal and ash, which can include bulldozing, rail delivery, and wind-driven lofting and advection. Road dust can also be present. Three sensor systems from different manufacturers were tested to see if they could detect dust plumes. All systems incorporated sensors using optical techniques to count particle number, with conversions required to produce results in units of particle mass. Two of the systems were powered with small solar panels and batteries, rather than through connection to grid power (see Figure 2). The sensor results will be compared against reference instruments, including hourly Federal Equivalent Method data, sub-hourly data, and particle size distribution data. This deployment was only recently completed, and analysis is ongoing, but early results suggest that some of the deployed sensors did capture many of the dust plumes of interest. Good precision was observed between duplicate sensors of the same manufacturer and model (see Figure 3). Accuracy is still to be determined. As the field study lasted for nine months, an initial consideration of the sensor lifespan and longterm replacement frequency is possible. Impacts of seasonal meteorology on sensor performance will also be investigated. EPRI is also testing sensors in a groundwater monitoring deployment using down-hole multi-sound sensors equipped with probes for ph, electrical conductance, chloride, temperature, and groundwater elevation. There is currently no sensor that can completely replace a traditional groundwater monitoring program, particularly for inorganic applications,
because probes are not commercially available for constituents such as boron and sulfate dissolved in groundwater. However, there may be applications where sensor information can supplement traditional monitoring using surrogate constituents such as electrical conductance and chloride. For example, providing higher time resolution information between sampling events may be of value in groundwater environments with rapid flow, such as karstic groundwater systems. Another potential use is monitoring of indicator constituents during remediation. For example, one use could be to optimize operation of an enhanced groundwater flushing system, where clean groundwater is injected at some wells and impacted groundwater is extracted at other wells, and where injection and pumping rates can be modified in response to system effectiveness as indicated by the sensor array. 1 Results of EPRI testing to date have demonstrated the ability of these instruments to achieve the objective of monitoring short-term fluctuations in groundwater quality. The results have also suggested that maintenance requirements for these devices, in their current stage of development, makes them best suited to specialized applications. make necessary adjustments in the field. This requires training of individuals to understand how to calibrate sensors and otherwise maintain the equipment. Important Study Design Features Best practices for study design and implementation learned from the EPRI deployments are evolving. Most practices relevant at utility sites were like those previously recommended for ambient sites. For example, it was important to test multiple sensor devices of same manufacturer and model, as well as multiple different models. This helped to detect and address Figure 2. Example of an air sensor node on a tripod stand, with solar panel and battery for power provision. Other innovative environmental sensor applications have also been tested. One utility member of EPRI has set up an extensive array of geotechnical sensors that provide real-time data to monitor dike stability at impoundments using vibrating wire transducers to monitor pore water pressure, inclinometers to measure lateral movement, and borehole extensometer to monitor settlement/vertical movement. EPRI is documenting this successful application so that other companies can consider similar deployments. 2 Both expected and unexpected challenges were encountered during these pilot deployments. For example, complications with the air particulate matter sensors were expected at high relative humidity levels, and were indeed observed. However, prior deployments and manufacturer specifications did not provide prior indication of the number of large false positive signals that were observed at temperatures below zero degrees Celsius. Additionally, an unexpected failure of electromagnetic compatibility was found during a deployment at a power generation facility, which was associated with a single component in a sensor system package. This same package design was deployed at a variety of other types of sites (e.g., rural, urban) with no indication of any issue. A substantial upfront labor investment was required for the pilot deployments to understand drivers of sensor performance and to determine likely maintenance needs. These labor costs could likely be reduced with future deployments, when a sensor network is beyond the pilot stage. In addition to upfront labor, resources are needed on an ongoing basis to check sensor readings for evidence of drift or malfunction, and to
Figure 3. Example field measurements in particles per cc of air for duplicate sensor systems of the same manufacturer and model (x- and y-axes) for (a) PM 1, (b) PM 2.5, and (c) PM 10 size fractions. issues with sensor stability, precision, and usefulness that would adversely affect results. Advance review of the technology specifications and laboratory testing cannot replace testing the devices of interest at the site of interest. It is crucial to include a collocated certified monitor (e.g., Federal Reference or Equivalent Method) or similarly vetted sampling and analysis protocols, as the best reference for comparison to sensor performance. This will help ensure accurate sensor readings and provide early indications of drift, confounding, or other issues. The reference instrument that provides data on a similar time interval as the sensor (i.e., typically less than 1 hour for air) enables the most robust comparison. Development of relevant data quality control and quality assurance approaches for long-term sensor operations and data management is also important. Ongoing intermittent comparison of sensor data against data from more traditional methods can be done on time periods suitable for the equipment and metrics being measured. Other details important to consider, from the perspective of an air sensor deployment, are summarized in a prior issue of EM. 3 For groundwater, another important consideration is whether the cost of installing and maintaining the sensor is lower than the cost of sending a person to collect and a laboratory to analyze the samples. Another consideration for groundwater applications is sample independence. In many non-karstic groundwater environments, multiple sampling events on the same day, or even same week, are essentially drawing from the same, slow-flowing, mass of groundwater. Therefore, chemical measurements are not independent and are not useful, which makes moot one of the advantages of using sensors, which is collecting many readings over a short period. This suite of details to consider during sensor deployments highlights the importance of incorporating a site-specific approach to project design. Another important component of any eventual sensor monitoring program determined to be suitable for real-world use at utilities is an alerting system. For such an application, data are accumulated, statistically evaluated or compared to metrics, and alerts sent when readings are out-of-bounds of expected values. Alerting facilitates the ability to take operational action to respond to the observations. Future Value All signs point to a continuation of research on environmental sensor applications and performance evaluations. EPRI will also continue to review new sensor developments and potential applications for opportunities and partnerships that may be relevant to electric utilities. These include applications relevant to ambient environments and traditional environmental monitoring networks, as well as facility monitoring. EPRI s focus is on technologies that provide high data quality while helping electric utilities to lower their monitoring costs, provide increased spatial or temporal density of measurements, or otherwise provide new insights to assist with managing facility operations. em Stephanie Shaw and Bruce Hensel are Principal Technical Leaders at the Electric Power Research Institute in Palo Alto, CA. References 1. Corrective Action Technology Profile: Groundwater Extraction and Treatment at Coal Combustion Residual Facilities; Technical Report 3002010945; EPRI, Palo Alto, CA: 2017. 2. SENTINEL: Geotechnical Instrumentation Overview Example Application and Cost. EPRI, Palo Alto, CA: in-press. 3. Dye, T.; Graham, A.; Hafner, H. Air Sensor Study Design Details Matter; EM November 2016.http://pubs.awma.org/flip/EM-Nov-2016/dye.pdf