In our last post, we went over the top three opportunities to improve the performance of reverse osmosis plants with machine learning, a form of applied artificial intelligence. The first of those three opportunities we covered is what we're expanding on in this post: membrane servicing.
Membrane servicing is a principal cost and performance driver for RO systems which has seen very little innovation in the last 30 years, according to the International Desalination Association (IDA). It's also a major part of the process. Since RO membranes reject such high percentages of suspended and dissolved solids, they require proper pretreatment and maintenance to prolong their working life and delay their eventual fouling.
Before discussing RO membrane servicing solutions, it's important to understand the limitations of today’s membrane health management practices.
Curious about the brains 🧠 behind the management of membranes? Click here to access our in-depth Reverse Osmosis Whitepaper
REACT - MEMBRANE SERVICING TODAY
Timely and effective membrane servicing is essential to maintain RO system performance and optimize membrane life, both of which save time, money, and the environment. Irrespective of pre-treatment units and RO system design, operators and engineers face daily challenges in determining optimal membrane servicing (1. To clean or replace membranes, 2. How to clean) and schedules (when to service) in an opportunity-cost balancing act of downtime versus forgone performance restoration. Current best practices for membrane servicing include both reactive and scheduled preventative procedures:
Reactive Procedures – service membranes when an asset fails or when a performance metric exceeds recommended limit (e.g., membrane OEM datasheets and operating philosophies). Common reactive procedures include but are not limited to the following:
Perform cleaning (flush or chemical cleaning) when normalized parameters fall below acceptable threshold (e.g., pressure drop increases 10-15%)
Determining optimal cleaning procedure (e.g. chemical type, temperature, pH) using a combination of tribal knowledge and membrane vendor guidelines
Conduct membrane autopsy analysis to diagnose poor membrane performance and guide future service activities
Preventative Procedures – prescheduled frequency-based service and maintenance activities to restore membrane performance. A sample checklist of preventative maintenance activities to ensure effective RO performance including membrane health is summarized below:
Figure 1 - General preventative maintenance schedule at an RO facility
These membrane service workflows require significant manual analysis paired with skilled operator judgment to determine the best course of membrane servicing. Membrane servicing (replacement and chemical costs not including labour) account for 21% of the total OPEX at a typical seawater reverse osmosis (SWRO) plant. Unfortunately, this is the price of being in business and many RO plants make do with these strategies to meet their product water requirements.
“Membrane servicing (replacement and chemical costs not including labour) account for 21% of the total OPEX at a typical seawater reverse osmosis (SWRO) plant.”
PREDICT - MEMBRANE FORECASTING
RO membrane service innovation represents an untapped opportunity for the treatment space. Pani’s AI Coach™, a web-based platform designed for RO treatment applications, seamlessly integrates with existing systems (SCADA, LIMS, and PLC systems) to centralize data and provide timely guidance on optimal membrane servicing (1. What to service, 2. How to service, 3. When to service) to extend effective membrane lifetime, prevent system upsets, and improve operations and maintenance (O&M) efficiencies.
One example of Pani’s Membrane Tools is Membrane Forecasting which predicts days in advance when performance criteria (e.g., normalized salt passage, product flow rate) or system limits (e.g., pump speed) will exceed safe thresholds (see figure 2). Pani’s predictive models mitigate guesswork and provide a planning tool for operators and plant managers to know the optimal time to service membrane assets, preventing higher chemical consumption from overcleaning and loss of membrane life due to excessive cleaning cycles:
Figure 2 - Pani Digital's RO forecast tool predicting and visualizing when product conductivity will exceed safe limits to help operators plan ahead
O&M staff and service professionals can be notified by text message or by e-mail when a membrane service is required.
SWRO USE CASE
Let's quantify these benefits with a use case. A large seawater reverse osmosis (SWRO) facility in California with an average daily capacity of 189 MLD provides clean drinking water to approximately 400,000 people. The desalination plant has strict treatment standards, with product salinity limits of 200ppm using a 2-pass RO process.
The facility has an annual OPEX of USD $53.1 million. Membrane replacements account for an estimated $4.6 million, and average unit chemical costs (sodium bisulfate and scale inhibitor for membrane cleaning) of $0.07/m3 for every unit of water produced.
Pani Digital’s analytical tools and recommendations can extend membrane life by 20-35% and reduce chemical consumption by 3-10%. This equates to $1.6M potential savings in membrane replacement and $386,000 chemical costs offset for this plant each year.