⚡ electrical engineering

Water- and energy-efficient dialysate flow optimisation in haemodialysis using a data-driven porous media model

This study introduces a data-driven porous media model to optimize haemodialysis by identifying dialysate flow rates that significantly reduce water and energy consumption while maintaining high solute clearance, thereby establishing a framework for more sustainable and patient-specific dialysis strategies.

Original authors: Ruhit Sinha, Anne Staples

Published 2026-09-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ruhit Sinha, Anne Staples

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

For millions of people around the world, life depends on a machine that acts as an artificial kidney. This treatment, known as haemodialysis, filters waste from the blood when the body's own kidneys can no longer do the job. The process works by pumping blood through a bundle of thousands of tiny hollow fibres, while a special cleaning fluid called dialysate flows around the outside of these fibres. Waste products move from the blood, through the fibre walls, and into the cleaning fluid, which is then discarded. While this therapy is a medical miracle, it comes with a heavy environmental cost. Each treatment requires a massive amount of highly purified water to create the cleaning fluid, and the pumps that move the blood and fluid consume significant electricity. A single four-hour session can use over one hundred litres of water, and when multiplied by the millions of patients treated globally, the resource demand becomes staggering.

The central challenge for engineers and doctors has been balancing the need for effective cleaning with the desire to use less water and energy. For decades, the standard practice has been to pump the cleaning fluid at a high rate, roughly twice the speed of the blood flow, to ensure toxins are removed quickly. However, this approach assumes that more water always means better cleaning, without fully accounting for the energy required to push that water through the machine. Researchers have long suspected that this high-flow method might be wasteful, but proving it required a way to predict exactly how the fluid moves through the complex maze of fibres and how that movement affects both cleaning power and energy use. Without a reliable way to model these hidden flows, doctors have been unable to prescribe lower water rates with confidence, fearing they might compromise patient safety.

A team of researchers at Virginia Tech has now developed a new way to look at this problem, creating a computer model that treats the bundle of fibres not as individual tubes, but as a single, porous sponge-like structure. By simulating the flow of fluid through twenty different types of dialysis machines, they discovered that the old mathematical rules used to predict how hard it is to push fluid through these bundles were significantly off. The traditional formulas, originally designed for packed columns of sand or gravel, failed to capture the unique way water moves around the tightly packed fibres in a dialyser. In many cases, these old rules overestimated how easily water could flow, leading to inaccurate predictions of the energy needed. To fix this, the researchers built a new, data-driven formula based on their detailed simulations. This new model accurately predicts the pressure needed to move the cleaning fluid, reducing the error in predictions from nearly ninety percent down to about forty-five percent.

With this more accurate model in hand, the team introduced a new way to measure the efficiency of dialysis: how much waste is removed for every unit of energy spent pumping the fluids. They called this metric "clearance per pumping power." When they applied this metric to a standard dialyser, they found that the most efficient operating point was not the high-flow setting used in most clinics today. Instead, the simulations showed that the best balance between cleaning power and energy use occurred when the cleaning fluid flowed at a rate closer to, or even slightly slower than, the blood flow. Specifically, for a patient with a blood flow of 300 millilitres per minute, the model suggested that reducing the cleaning fluid flow from the standard 500 millilitres per minute down to 280 millilitres per minute would save nearly fifty litres of water per session. Remarkably, this reduction would still remove about ninety-three percent of the urea, a key waste product, that is removed at the higher flow rate. To make up for the slight difference in cleaning speed, the treatment would need to last only about eighteen minutes longer, a small trade-off for such a large saving in water.

The researchers also explored how this approach might play out on a global scale. By combining their findings with data on where kidney disease is most common and where fresh water is scarce, they created a map showing which regions could benefit most from switching to these optimized, lower-flow settings. The results highlight that countries in the Middle East, North Africa, and parts of South Asia, where water is already a critical resource, could see the most significant environmental relief. In these areas, the cumulative water saved by millions of patients switching to more efficient flow rates could amount to billions of litres annually. While these findings are currently based on computer simulations and have not yet been tested in a clinical trial, they provide a strong theoretical foundation for rethinking how dialysis is prescribed. The study suggests that by understanding the physics of fluid flow more deeply, it is possible to design treatment strategies that are not only effective for patients but also sustainable for the planet. The framework developed here offers a path forward for doctors to consider water and energy efficiency alongside medical outcomes, potentially transforming a life-saving treatment into a more environmentally friendly one.

Technical Summary: Water- and Energy-Efficient Dialysate Flow Optimisation in Haemodialysis Using a Data-Driven Porous Media Model

Problem Statement
Haemodialysis imposes a substantial environmental burden due to high water consumption for dialysate production and significant electricity usage for hydraulic pumping. While clinical studies suggest that reducing dialysate flow rates ( $Q_{Di}$ ) can maintain dialysis adequacy (measured by $Kt/V$ ), current clinical prescriptions do not explicitly account for the hydraulic power required to achieve specific solute clearances. Existing computational models for dialyser hydrodynamics rely on single-fibre models or classical porous media correlations (such as the Ergun equation), which the authors identify as exhibiting systematic errors when applied to the complex geometry of hollow-fibre membrane modules. Consequently, there is a lack of a framework that couples toxin clearance with hydraulic energy expenditure to identify water- and pumping-energy-efficient operating conditions.

Methodology
The study employs a multi-scale computational approach to develop and validate a data-driven porous media model:

3D Navier-Stokes Simulations: The authors performed module-scale three-dimensional Navier-Stokes (3D N-S) simulations for 20 different dialyser geometries with porosities ranging from 0.4 to 0.8. These simulations served as the ground truth for hydrodynamic behavior.
Model Derivation: By analyzing the 3D N-S results, the authors identified semi-empirical relationships governing permeability in hollow-fibre bundles. They derived a modified Ergun equation that incorporates:
- An additional transitional inertial term proportional to $Re'^{0.9981}$ .
- Revised axial permeability expressions including a wall-effect parameter ( $M$ ) and a modified Blake–Kozeny constant ( $\alpha_{zz}$ ).
- A semi-empirical relation for the inertial resistance term based on axial permeability ( $\kappa_{zz}$ ).
- A fixed fractional relationship for radial permeability ( $\kappa_{rr} \approx 0.51 \kappa_{zz}$ ).
Validation: The modified model was calibrated against the 3D N-S data and validated against published experimental pressure-drop measurements from 11 hollow-fibre dialyser modules (including PHYLTER® and UNIPA test modules).
Performance Metric (CPP): The authors introduced Clearance per Pumping Power (CPP), defined as the volume of blood cleared of a toxin per unit of hydraulic pumping power. This metric was used to optimize operating points for a modelled PHYLTER® HF 17SD dialyser.
Solute Transport: A double-porous media model coupled with convection–diffusion equations was used to predict toxin clearance (urea and vitamin B12) under various flow conditions, utilizing the modified Ergun equation for dialysate-side pressure drop.
Global Impact Assessment: A "Dialysis Water Savings Potential" score was developed to estimate country-level water savings based on chronic kidney disease (CKD) prevalence and renewable freshwater resources, assuming a shift from conventional flow rates to CPP-optimized rates.

Key Results

Model Accuracy: Single-fibre models and the classical Ergun equation systematically overpredict or underpredict permeability and pressure drop. The modified Ergun equation reduced the mean absolute percentage error in pressure-drop predictions relative to experimental data from approximately 90% (classical) to 45%.
Optimal Flow Ratios: For the PHYLTER® HF 17SD dialyser, CPP is maximized at dialysate-to-blood flow ratios ( $Q_{Di}/Q_{Bi}$ ) of approximately 1.1 to 0.7, depending on the solute and ultrafiltration rate. This is significantly lower than the conventional ratio of ~2.0 or the 1.25–1.5 range often used in adaptive systems.
Water and Energy Savings: At a blood flow rate ( $Q_{Bi}$ ) of 300 mL min⁻¹ and ultrafiltration rate ( $Q_F$ ) of 30 mL min⁻¹, reducing the dialysate flow from 500 to 280 mL min⁻¹ saves approximately 48 L of water per session while retaining 93% of urea clearance. To maintain equivalent dialysis adequacy ( $Kt/V$ ), the session duration would need to increase by only ~18 minutes.
Global Potential: The study projects that optimizing flow rates could save tens of liters per patient per session. Aggregated globally, this represents billions of liters of potential water savings, with the highest relative impact in water-scarce regions like the Middle East, North Africa, and South Asia.

Significance and Claims
The paper establishes a mechanics-based framework for evaluating the trade-offs between solute clearance and hydraulic pumping energy in hollow-fibre membrane modules. The authors claim that:

Hydrodynamic Modeling: The modified Ergun equation provides a substantially improved reduced-order estimate for dialysate-side pressure loss, addressing the geometric limitations of classical correlations in hollow-fibre systems.
Operational Optimization: The CPP metric identifies device- and solute-specific operating points that maximize efficiency, demonstrating that conventional dialysis parameters are not energetically optimal.
Sustainability Strategy: The framework supports the development of water- and energy-efficient dialysis strategies. While the authors note that clinical feasibility requires further validation (citing preliminary studies on lower flow rates), the model provides testable hypotheses for device-specific and patient-specific prescriptions.
Generalizability: The approach of coupling transport performance with hydraulic energy expenditure is presented as a general mechanical engineering framework applicable to other hollow-fibre systems in water treatment, desalination, and medical devices (e.g., extracorporeal membrane oxygenators).

The study concludes that while it does not capture whole-machine electricity consumption or patient-specific safety constraints, the identified operating points and country-level water savings scores offer valuable design-screening outputs for advancing sustainable haemodialysis.

More like this