Article Text


Original article
Indigenous ultra low cost technology for spinning haemodialysis grade hollow fibre membranes
  1. Anirban Roy1,
  2. Lloyd Vincent2,
  3. Sirshendu De1
  1. 1Department of Chemical Engineering, Indian Institute of Technology, Kharagpur West Bengal, India
  2. 2Clinical Director Dialysis Services at Narayana Hrudayalaya Hospital, Bangalore, Karnataka, India
  1. Correspondence to Dr Sirshendu de, Department of Chemical Engineering, Indian Institute of Technology, Annexe Building, Kharagpur, West Bengal, P.O. Box 721302, India; sde{at}


Background The present paper is about developing an indigenous technology to spin ultra low-cost hemodialysis fibres. It presents a complete engineering design and specific operating conditions to spin dialysis grade, hollow fibre membranes.

Methods Simple, lab-made, disposable needle assemblies have been used to extrude fibres, replacing the conventional spinneret-based technique. A complete arrangement of spinning paraphernalia has been designed and discussed to make the process less energy intensive. Biocompatible polymer such as polysulfone (PSf) has been used as base material, along with biocompatible additives such as polyvinylpyrrolidone (PVP) and polyethylene glycol (PEG-200).

Results Three grades of hollow fibres have been spun (6, 12 and 16 kDa molecular weight cut-off) and are characterised in terms of surface morphology and molecular weight cut-off. In vitro clearances and the transient permeation of uremic toxins has also been evaluated in diffusion governed dialysis mode.

Conclusions An ultra low cost technology was developed and optimised to spin dialysis fibres of clinical specifications, with inner diameter of 220 µ and thickness of 35–40 μ. 6 kDa fibres are found to be high efficiency dialysis grade, whereas 12 and 16 kDa are potential high performance membranes.

  • Affordable
  • Frugal
  • Inventions
  • Assistive Technology

Statistics from


Kidney failure is quite common nowadays, due to the genetic anomaly, or lifestyle adopted by a person. Haemodialysis (HD), or peritoneal dialysis (PD) is the only life support system for a kidney failure patient. Glomerular filtration rate (GFR) of a person with a properly functioning kidney is around 120 mL/min.1 GFR is a quantitative indicator of kidney's functioning ability and its value below 20 mL/min indicates the patient has severely impaired kidney function,1 requiring dialysis. However, a patient is seldom subjected to PD, since it is not suitable for everyone. Moreover, social constraints limit its applicability2 making HD preferred mode of treatment for kidney patients. However, HD can prove to be a very expensive treatment procedure for an average family, and cost is mainly due to the dialysis cartridge. An average patient requires 144 cartridges a year, resulting in huge costs due to cartridges alone. This amounts to around 58% of the total cost per session of dialysis.3

Hollow fiber dialysers contain clinically specified, 7000–15 000 specialty membranes, which are spun by spinneret-based technology. Custom made spinnerets4 ,5 as well as capillary channels6 have also been used to spin dialysis grade hollow fibres. However, such custom made spinnerets are expensive and dialysis fibres of mentioned dimensions are not easy to attain due to proprietary technology. In case of a developing economy such as India, where 70% of the population is below poverty line,7 imported expensive dialysers become unaffordable for most.


The present article is a novel effort undertaken to simulate the perfect spinning conditions to obtain clinical grade hollow fibres utilising a cost-effective technology. The most important aspect is that extrusion of fibres has been achieved by custom-made disposable syringe assembly and optimisation of the process conditions in a less energy intensive method.8 The developed technology can be the real future to provide affordable renal treatment to dialysis patients, especially in developing countries across the globe. This article first introduces the concept of spinning hollow fibres using syringe-based assembly and the details of the process. The cost of extrusion mechanism, designed for spinning of the hollow fibres, is around $0.3 (₹20). Then it discusses the fine tuning of such developed fibres by varying the composition of the polymer solution so as to obtain fibres of different molecular weight cut-off (MWCO) used for HD purposes. Lastly, the detailed characterisation and in vitro performance of the developed hollow fibres and their capability to reject uraemic toxins have been investigated.

Materials and methods


Polysulfone (PSf), the base material of the membrane was obtained from M/s Solvay Chemicals, Mumbai, India. Polyvinylpyrrolidone (PVP) (molecular weight 44 000) was obtained from M/s Sigma Aldrich and polyethylene glycol (PEG) (molecular weight 200) was supplied by M/s S R Ltd, Mumbai, India. Different molecular weight of neutral polymers of PEG ((4, 6, 10, 20, 35 and 100 kDa), creatinine and sucrose were procured from M/s S R Ltd. The solvent, di-methyl formamide (DMF) and urea were obtained from M/s Merck (India), Mumbai, India. The syringes and adhesives for spinning and joining of the needles, respectively, were procured from the local market. Bovine serum albumin (BSA) was obtained from Himedia, Nasik, India. Vitamin B12 was procured from M/s Sigma Aldrich.

Membrane synthesis

Solution preparation

In total, 18 wt% PSf, 0 wt% PVP and 3 wt% PEG-200 were dissolved in DMF and was stirred with slow heating at 60°C. The solution was kept on stirring until the polymer was dissolved completely, yielding a solution with golden colour. It was sealed and kept overnight for degassing. Three types of hollow fibres were spun in this work with PVP concentration of 0, 2 and 3 wt%, respectively.

Membrane spinning

Membrane spinning is the pivotal part of this particular article and is discussed in detail in a separate section.

Molecular weight cut-off

The MWCO was measured by measuring the rejection of neutral polymers (PEG) of various molecular weight, such as, 2, 4, 6, 10, 35 and 100 kDa. A solution of 10 kg/m3 of each polymer was filtered using a transmembrane pressure drop of 10 mm Hg and cross flow rate of 10 mL/min keeping the dialysate circuit open. An Abbe-type refractometer, supplied by M/s Excel International, Kolkata, India, was used for the measurement. The observed rejection (% Robs) was calculated as:Embedded Image 1where, CP and CF are the concentrations of the permeate (kg/m3) and feed (kg/m3), respectively.

Surface morphology

The surface morphology of the hollow fibre membranes were studied using a Scanning Electron Microscope (SEM) (model: ESM-5800, JEOL, Japan). The membranes were dried overnight in a desiccator. The membranes were cut to proper size and were gold coated and placed on stubs for the cross-section and thickness images.

Other membrane characterisation and in vitro testing

The membranes were characterised in terms of their hydraulic permeability, contact angle, breaking stress, porosity and pore density. The in vitro tests, such as diffusive permeability, clearances, k0A and Kt/V were also conducted and their methodology is presented in detail in online supplementary material.

Scaling of the experiments

Since number of fibres was used as 300, the flow rates were reduced, and the clearance, k0A and Kt/V values obtained from the experiments were then scaled up to the actual cartridge, with filtration area of 1 m2 and flow rates in the order of 250 mL/min. As discussed earlier, clearance is expressed by eq. (7) (see online supplementary material).

Reynolds number (Re) in each fibre is defined as:Embedded Image 2where, ρ is the density of blood side solution, µ its viscosity (Pa s), n the number of fibres, R the inner radius (m) of fibre, Embedded Image the blood side fluid flow rate (mL/min).

Thus, eq. (2) is expressed in terms of Re as:Embedded Image 3where, Embedded Image, Embedded Image is solute clearance (mL/min), Embedded Image the ratio in eq. (3), Embedded Image the inlet solute concentration (mg/L) and Embedded Image the outlet solute concentration (mg/L).

Keeping Re through one fibre same in two different cartridges (1 and 2), and all other physical constants remaining same, the scaling rule becomes:Embedded Image 4

Embedded Image was measured experimentally with n1=300. Embedded Image for a commercial cartridge, with flow rate 250 mL/min and 7500 (Embedded Image) fibres (effective filtration area 1 m2), is 0.6.9 Therefore, clearance (Embedded Image) is calculated using eq. (4) for a cartridge with 1 m2 filtration area and 250 mL/min for flow rate.

Membrane spinning

This section details the in-house membrane spinning. Membranes are formed when a polymeric solution comes in contact with an antisolvent such as water (also referred to as gelation bath) and mass transfer occurs. In figure 1A, a simplistic process description is presented. The yellow layer is polymeric film, which when comes in contact with water (blue), the solvent molecules are transported out from the film and water gets transported in (figure 1B). This creates pores in the structure resulting in formation of membrane. The mechanism is similar in case of hollow fibre membranes, where, a central core of water is surrounded by concentrically flowing polymer solution. The mass transfer occurs radially creating pores (figure 1C). This basic principle of membrane formation is captured in design of spinnerets. Figure 1D depicts the cross-section of a commercial spinneret. Here, the central bore has water flowing through it and the bores on either side has polymeric solutions. The polymeric solution as well as water are pumped and dimensions of bores, individual flow rates and pumping rates are all part of spinneret design and a delicate balance among all ensures proper spinning of fibres. All these parameters, combined, determine the adherence to the clinical dimensions of 180–220 and 35–40 µ thickness.

Figure 1

(A) Polymer film-water system, (B) mass transfer and pore formation in a polymer film-water system, (C) mass transfer mechanism in a hollow fibre membrane, (D) commercial spinneret cross-section and flow pattern of a polymer non-solvent, (E) basic design of syringe-in-syringe (SIS) assembly and (F) Preliminary spinning set-up.10

In this present research effort, the basic operation principle of spinnerets has been understood to mimic their functioning. The authors have resorted to use of syringes to simulate the flow patterns of polymer-water in conventional spinnerets. Figure 1E shows a laboratory fabricated syringe-in-syringe (SIS) assembly diagram where, a smaller diameter syringe has been inserted into a larger diameter syringe. The larger diameter syringe has a hole drilled on the lateral surface and the smaller diameter syringe is bent and inserted into it. The authors have started the investigation with a range of syringes and operating parameters. The set up designed for this purpose was initially described in Thakur and De10 and is presented in figure 1F. The polymer flow is facilitated by pressure from nitrogen cylinder and water flow was facilitated by a booster pump. The Polymer flows through the SIS assembly's annular section (figure 1E, blue) and water through the inner diameter syringe (black). The hollow fibre membranes are extruded and fall into the gelation bath (figure 1F) and are then wound on a spool. This arrangement was used as starting point for the HD membrane spinning exercise. The authors started with SIS assemblies consisting of 16 gauge (16G) outer syringe and 24G inner syringe. The fibres extruded from this was unacceptable for dialysis membranes and has not been shown in this article. This led to use of syringes with smaller diameter and trials with flow rates. Figure 2A depicts the use of various combinations and scanning electron microscope images of the membranes obtained. Figure 2A–I depicts an 18G outer syringe combination with U-100 insulin syringe. A pump was used for the extrusion, but the membrane had 300 µm thickness and diameter of 240 µm. This was unsuitable for dialysis and hence smaller outer syringes were used with no use of pump, with water being driven entirely due to gravity. A 19G with U-100 SIS combination, 20G and U-100 combinations were also tried. The thicknesses of membranes were reduced but inner diameter was above 220 µm. Finally, after a lot of iterations with flow rate and using a 22G-BD32 gauge insulin syringe combination, was a dialysis membrane finally obtained with 220 µm inner diameter and 35 µm thickness. The final detailed design is discussed below.

Figure 2

(A) Needle combinations and Scanning Electron Microscope (SEM) images of hollow fibre membranes. (B) Syringe assembly: (a) disposable syringe and (b) insulin syringe. (C) Detailed drawing of the polymer and water cylinders: (a) polymer tank, (b) water tank, (c) ball valve, (d) needle valve and (e) syringe assembly. (D) Schematic diagram of the spinning set–up: (a) nitrogen cylinder for pressurising, (b) bucket, (c) steel structure, (d) gelation bath and (e) spool for winding the fibres. All dimensions are in centimetres.

This is depicted in figure 2B. A syringe of 230 µ diameter was bent at angle of 120° and inserted into an outer needle of diameter 700 µ. Both the syringes were commercially available, each costing ₹ 10 (US$0.15). This assembly was sealed with adhesive to make it leak proof and also to maintain the concentric arrangement of the needles, as depicted in figure 2B. There were two feed tanks for polymer solution and water. The polymer flow was controlled by ball valve and the water flow, requiring finer control, was regulated by a needle valve. The syringe assembly (marked e) was joined to the water tank by a tube at right angles and the water flowed through the syringe and the polymer flowed through the annular space. The distance maintained between the cylinders was 12.5 cm. The difference in height between flanges was 9.5 cm and the polymer tank was 4 cm lower than the water tank. This unit was the fundamental subunit of the complete set up illustrated in figure 2D. The polymer flow was facilitated by the pressure applied (140–200 kPa) by the nitrogen cylinder and the water was flown from an overhead tank due to gravity. The minimum height maintained for the spinning of fibres was 160 cm from the ground. The water flowing through the syringe formed the hollow core of the fibre which, after extrusion, was put into the gelation bath containing normal tap water. The fibres were then wound on the motor-driven spool. In this whole arrangement, there was no need of electricity anywhere, except in the final step where the spool was rotated by a voltage variable frequency drive. Cost of syringe assembly was $0.3 (₹ 20) and the rest of the paraphernalia too were non-expensive making this technology to be ultra-low cost. The specific conditions for the spinning were tabulated in table 1.

Table 1

Conditions for spinning of dialysis fibres

Results and discussions

MWCO and surface morphology are membranes along with their clearance and urea and creatinine transient transport capabilities is discussed in MWCO and surface morphology section. Other characterisation and performance parameters are discussed in detail in online supplementary material.

MWCO and surface morphology

MWCO of various membranes is presented in figure 3. Three kinds of dialysis fibres in the range 6–16 kDa were obtained depending on PVP concentration. PVP induces biocompatibility11 whereas PEG reduces the pore size of membranes.12 It was found that with PVP of 0 wt%, PSf 18 wt% and PEG 3 wt%, a 6 kDa membrane was obtained. Increased PVP concentration 2 and 3 wt% resulted an increase in MWCO to 12 and 16 kDa, respectively. This is explained based on phase inversion mechanism. The mechanism of phase inversion characteristic of PSf-PVP-PEG membrane is discussed in literature.12 ,13 It can be inferred that dense skin is due to the presence of PEG and macropores in the membrane structure is due to presence of PVP. PVP is a hydrophilic polymer and adding it to a base polymer such as polysulfone, increases the hydrophilicity of the resulting membrane. While phase inversion, the presence of PVP enhances the mass transfer rate of water (antisolvent) into the polymer matrix, thereby, leading to the formation of macrovoids. In other words, PVP leads to quick demixing of the polymeric solution during phase inversion and absence of PVP leads to delayed demixing. Delayed demixing gives rise to a dense structure, whereas, quick demixing gives rise to formation of macrovoids. This can be seen from SEM images, that membrane with 0% PVP has a dense structure, but as percentage of PVP is increased, the non-solvent flux into the matrix increases thereby increasing macrovoids. This can be verified by SEM images (figure 4). It is evident that macrovoids increase with PVP concentration. A close look at cross-sections of figure 4 reveals 0% PVP yields a dense morphology in the membranes (figure 4A, B). The porous substructure becomes more pronounced and more macrovoids are formed at higher PVP concentration resulting in increase of MWCO of the membranes from 6 to 16 kDa. It is reported in literature that high performance membrane (HPM) synthesised by Rhone Poulenc (AN 69) has pore size of 29 Å14 whereas polyamide membranes (Gambro) have externally macroporous pore sizes of 40–50 Å.14 The 6 kDa membrane has an average pore size of 30 Å, whereas, 12 and 16 kDa membranes in present study have pore sizes of 40–42 Å, as seen in online supplementary figure S3(b). However, the cross-section of membranes spun from this technology is little non-uniform. This is observed from the SEM images. It has to be understood that the SIS assembly constructed has a role to play in uniformity of the cross-section. However, this can be circumvented lending a more professional finish to the prototype designed.

Figure 3

Molecular weight cut-off (MWCO) of the dialysis grade membranes.

Figure 4

Scanning Electron Microscope (SEM) images of the dialysis membranes: cross-section (a) 6 kDa, no polyvinylpyrrolidone (PVP); (b) 12 kDa, 2% PVP; (c) 16 kDa, 3% PVP; thickness (d) 6 kDa, no PVP; (e) 12 kDa, 2% PVP; (f) 16 kDa, 3% PVP.

Diffusive clearance

In an actual HD, a transmembrane pressure (TMP) is applied across membrane leading to ultrafiltration, in order to enhance solute clearance. This mode of dialysis is known as haemofiltration or haemodiafltration.15 β2-Microglobulin (B2M) removal is essential during dialysis apart from urea and creatinine. Such dialysers are known as ‘High Performance Membrane Dialyzers’ (or HPM) and advantages and transport properties of these are well documented.9 ,16–18 High-performance membranes have impacted the dialysis scenario to a great deal. It has been observed that membranes with higher pore sizes mimic the pores of glomerulus (30–50 kDa).19 The larger pore-sized membranes result in loss of albumin during dialysis (<3 g/session).20 The Japanese Society of Dialysis Therapy advocates the loss of albumin, since this leads to more loss of B2M, inflammatory cytokines accompanied with huge removal of urea and creatinine.21 In fact, HPMs have made it possible to achieve higher removal rates of middle molecules in shorter dialysis durations. Moreover, it was also reported that mortality and hospitalisation rates were lowered with usage of synthetic high-flux membranes.20 From clearance values, it can be inferred that 12 and 16 kDa membranes are potential HPMs and 6 kDa is a conventional high-efficiency dialysis membrane. Figure 5A shows that clearance increases with feed flow rate and highest clearance is realised at 300 mL/min. Urea and creatinine clearances at this flow rates is 200 mL/min for 12 and 16 kDa and 180 mL/min for 6 kDa. As expected, creatinine clearance is slightly lower than urea clearance due to its larger molecular weight. Creatinine clearances at 300 mL/min of feed flow rate is around 165–180 mL/min for the 12 and 16 kDa membranes and 150 mL/min for the 6 kDa fibres. In the reported literature, clearances of high-performance dialysers are in the range 200–300 mL/min. However, this is based on two aspects: (1) optimum design of dialysers, that is, packing efficiency, inclusion of spacers and, most importantly, (2) convective mode of transport. These two effects make the transport of uraemic toxins higher than what is reported in this paper. This work is about spinning of fibres and rudimentary packing with in vitro characterisation of membranes in primarily diffusion mode. This leads to clearance values less than what are reported.

Figure 5

(A) Diffusive clearance values of urea and creatinine. (B) Effect of feed flow and dialysate flow rates on the urea and creatinine transport through the membranes.

Effect of feed and dialysate flow rates

The ratio of Feed Flow Rate (FFR):Dialysate Flow Rate (DFR) has been maintained at 1:1 and 1:2. In state-of-the-art dialysis sessions, the blood flow rates are kept high in the range of 300 mL/min. Any flow rate beyond this starts an unusual build-up of pressure increasing the ultrafiltration rate of fluids across to the dialysate side, which can prove to be life-threatening. Once the blood flow is fixed, the dialysate is maintained approximately at 1:2, that is, 600 mL/min.22 In fact, the standard values are between 500 and 800 mL/min,22 but it is well reported that a 60% increase in the dialysate flow rate has a negligible effect on clearance (10–15% increase only).23 This has a negative impact as far as dialysate fluid consumption is concerned, thereby increasing the cost of dialysis. As observed from figure 5B (a) and (b), urea concentration is in decreasing order of 16, 12 and 6 kDa. Since DFR is increased to twice of FFR, the decline is faster due to the enhanced driving force of mass transfer.

It is observed that for FFR:DFR 1:1, a 6 kDa membrane needs 250 min to reach the recommended concentration (400 mg/L), whereas a 12 kDa membrane needs 175 min and a 16 kDa membrane needs around 120 min. However, when FFR:DFR is 1:2, this time drops to 200 min for the 6 kDa membrane and 120 min for both the 12 and 16 kDa membranes. It can be inferred that there is marked improvement as far as performance of the 16 kDa membrane is concerned over that of the 6 kDa membrane. However, the 12 and 16 kDa membranes have comparable performance. Similarly for creatinine, figure 5B reveals that for the FFR:DFR ratio of 1:1, the 6 kDa membrane takes 300 min to reach the 12 mg/L level, whereas the 12 and 16 kDa membranes take 250 min to reach the level. However, for FFR:DFR 1:2, the fall is more rapid and the desired levels are reached within 175–250 min.

In the present work, experiments are conducted in a purely diffusion-governed mode. In addition, the concentration gradient, that is, driving force across membrane, decreases with time as both the feed and dialysate are recycled. Therefore, the dialyser in this work performs poorly compared to a continuous supply of fresh dialysate in actual dialysis. However, even in this mode of operation, the dialyser with low-cost spinning technology is able to reduce the level of uraemic toxins to the desired limit in adequate time.

Scaling of the technology to plant scale

A feasible business model was developed and a detailed financial analysis was carried out. Various factors were considered to calculate the cost and price of a dialyser cartridge. The start-up fund requirements of the pilot plant are presented in table 2.

Table 2

Start-up fund requirements (INR)

A typical Special Economic Zone (SEZ) piece of land with an area of 2500 m2 was the basis of the calculation. The plant establishment's cost is ₹ 7 000 000 as per SEZ norms. The layout of the plot is divided into various activity zones such as shipping area, raw material input, fibre spinning area, potting area, etc. The plant production capacity is envisaged to be 34 000 cartridges per month, ramped up over a period of 3 years. On the basis of this and the process design as discussed in this work, about 3400 extrusion heads would be required to match the capacity, working at 16 h/day for a month. The various operating heads in the analysis are accounting, advertising, insurance, legal professional payment, license renewal, salaries, utilities, travel, telephone bills, meals and entertainment and website maintenance. Along with these, other operating expenditures include loan interest repayment, loan principal repayment and income tax. On the basis of these inputs and revenue generated from sales, using online ( software it is calculated that the cost of manufacturing one cartridge would be ₹ 168 and hence taking distribution into account, ₹ 300 would be the ultimate selling price of the same.

The comparative analysis between the commercial fibres and the developed ones is presented in table 3.

Table 3

Comparative analysis between commercial and developed fibres

It may be observed from table 3 that the developed fibres have the same desired clinical specifications compared to the commercially available ones, and can be developed at a much lower price using the technology presented herein.


A low-cost, syringe-based dialysis fibre spinning technique has been invented and the complete, detailed engineering has been discussed and the spun fibres have been characterised comprehensively. Fibres matched the clinical specifications of the inner diameter of 180–220 µ and thickness of 35–40 µ.

Three different MWCO membranes, 6, 12 and 16 kDa, were obtained by fine-tuning the polymer composition. Of these, 6 kDa fibres were of high efficiency dialysis grade and 12 and 16 kDa fibres were of high performance grade.

The performance of the three membranes for clearance was quantified and was found to be comparable with that in the reported literature. The urea clearance for the 6 kDa membrane was 180 mL/min, and for the 16 kDa membrane it was 200 mL/min for a feed flow rate of 300 mL/min.


The authors acknowledge the help of Kallol Paul.


View Abstract


  • Contributors AR was responsible for designing the experimental set-up and conducting experiments. Sd contributed to conceptualisation, designing of spinneret for extrusion of hollow fibres, and also contributed to polymer composition and optimisation of operating conditions for spinning. LV was responsible for designing and setting of specification of hollow fibres and quality monitoring.

  • Funding This work is partially supported by a grant from the Department of Science and Technology, Government of India, and Renalyx Health Systems Pvt. Ltd., Bangalore, India under scheme number IDP/MED/7/2011, dated 05.03.2012.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.