Designing Oxygen Concentrators for Low Resource Settings: Part 1 Resilience to Dust
This is the first part of a three part series looking at some of the lessons we have learnt relating to the design of oxygen concentrators in low resource settings.
Part 2: Designing Oxygen Concentrators for Low Resource Settings: Part 2 Resilience to Moisture
Introduction
Oxygen concentrators have proven an effective means of delivering oxygen therapy and are less expensive than cylinders [1,2]. However, oxygen concentrators were designed to operate in clean climate controlled settings, but in low resource settings many clinical hospitals have less than ideal environmental conditions — levels of dust and humidity can be higher than Western hospitals. Health services in low resource settings are extremely under-resourced and face some huge challenges. According to a World Health Organisation (WHO) lead review, approximately only one third of hospitals and clinics in Sub-Saharan African Countries have a reliable power supply [3,4], — meaning in these places, it is difficult to even run an oxygen concentrator.
In the summer of 2020, in partnership with Strathclyde University and Community Energy Malawi, we conducted a survey of a number of hospitals in Northern Malawi in order to try and understand the challenge firsthand. A summary of the findings is presented here. As we started talking to other stakeholders across the sector we realised there were similarities with what we were discovering in Malawi. We have heard many stories from around the world of the failure of oxygen concentrators and therefore the inability to reliably provide vital oxygen therapy. For example, oxygen concentrators breaking down and never being fixed, concentrators that had been stored in a warehouse not performing when they were needed, devices that just pump out air at room concentration due to failed sieve beds, oxygen concentrators caked in dust and not functioning. We have also learned about many oxygen concentrator graveyards (e.g. see Figure 1) and the work of OpenO2 to try and reverse this trend in Malawi [5]. After hearing these stories and collecting some data firsthand, we began asking whether a more resilient oxygen concentrator could be designed that would be able to withstand some of the harsher environmental conditions and be more appropriate for low resource settings, essentially a more robust oxygen concentrator that was designed to last. In this series of three articles we want to share some of the things that we have learned through the process of listening, designing, iterating and testing.
So why do they fail and what can be done?
There are a number of different reasons why concentrators break down and are not fixed. We found that clinical staff in Malawi had a limited expectation on the lifetime of concentrators, with reliability cited as a problem. So what kind of things are going wrong? Some of the causes of failure were actually very simple — faults with flow metres and power plugs and power cords were one of the leading causes of failure, or even just missing power cables. These kinds of problems are relatively easy to fix and so this points towards another problem — lack of maintenance and the challenges and issues around this. Other problems were more serious and more difficult to resolve quickly, for example failure of the sieve beds by moisture or failure of the compressor due to overheating caused by dust affecting cooling. Detailed reports on how/why oxygen concentrators break and how they can be fixed are available from OpenO2 in Malawi (https://www.openo2.org). From our research we have found that power related issues (surges, sags and power outages), dust and moisture were major factors that negatively affect the life and performance of oxygen concentrators used in low resource settings; a design overhaul is needed taking into account these aspects in order to make devices more fit for purpose. In addition, the majority of existing oxygen concentrator devices have very high power consumption — reducing this could improve their usefulness, especially in areas with limited power availability.
In the next section we are going to explore resilience to dust, and then in two following articles we will look at the challenge of moisture and aspects relating to power protection and energy efficiency.
Resilience to Dust
The dusty environment in sub-Saharan Africa has caused many oxygen concentrators to stop working properly. One of the most common reasons reported for this is blocking of the intake filter and failure to change it — a simple and relatively low-cost and low-skill maintenance task. Most oxygen concentrators have what is called a “gross particle filter” that will need to be changed regularly. If it is not changed, over time the increase in dust reduces the air flow, making the compressor work harder and increasing the pressure drop across the filter. This not only impairs the performance of the oxygen concentrator, it can cause the compressor to work harder and run at a higher temperature, reducing its life, and increasing the likelihood of failure.
Dust has been known to cause other problems too: it is abrasive, interferes with mechanics (such as the valves, e.g. the valves that do the cycling of the system often have fine tolerances and are susceptible to sticking if dust builds up). Dust can also cause circuit boards to fail. When a valve or circuit board fails, this can quickly lead to other problems, for example, it might lead to one bed being fed by room air continuously for an extended period introducing more water into one of the sieve beds, ultimately leading to sieve bed failure.
How do current oxygen concentrators manage dust?
Typically oxygen concentrators feature a cabinet air filter, an intake filter, and a final bacteria filter (see Figure 2 below for an example of the filters in the Devilbiss 525). The filters meet the medical requirements, however they can clog easily in dusty environments, especially when there are poor maintenance schedules in place. The cabinet filter — which prevents gross particles entering the system can be washed and replaced — but this does not keep out fine dust, rather it only prevents large particles, and things like pet hair from entering the system. However, the intake filter is more difficult to clean and it is recommended in the Devilbris manual that they are replaced with a like for like (i.e. the one made by the original manufacturer); it is this that can get blocked if not replaced.
One of the big issues is that product casing does not have an airtight seal and dust can enter at the bottom, drawn in by the compressor cooling fan. Figures 3 and 4, show how air, and therefore dust, can enter into the cabinet on both the DeVilbiss 525 and the Airsep VisionAire, respectively, two market leading brands. Dust can then circulate inside the product casing, and potentially interfere with the electronics and damage components — in particular the valves and control printed circuit board (PCB), as aforementioned. Potential failure modes of traditional oxygen concentrators as a result of dust, their effects and possible causes, have been identified and are summarised in Table 1 below (based on findings from Canta, Airsep and Devilbris devices).
Simple issues, such as build up of dust and lack of filter replacement, have been shown to lead to the end of a concentrator’s useful life. We asked the question — how might we increase the resilience of an oxygen concentrator to dust and reduce the mean time to failure. It is clear from the failure mode effect table above that the primary causes of the issues with dust are:
- the internal compressor filter becoming blocked;
- dust getting into and circulating within the oxygen concentrator itself.
It is important to be aware that in many leading oxygen concentrators on the market the cooling air intake for the compressor is not filtered air, and this is the primary entry point for dust into the system. Approximately 90% of the air used by an oxygen concentrator is for cooling the internal components (mainly the compressor).
We explored whether we would be able to:
- filter all of the incoming air going into the concentrator, this limits dust getting onto the components, the compressor or into valves;
- draw all air in from the top of the concentrator system, rather than from underneath the concentrator, to reduce dust being drawn into the system if it is being used on a dusty floor.
- protect the PCB and power management equipment such that even if dust makes it into the system these components have another layer of protection.
Developing a dust protection system of this type keeps the internal cabinet of the concentrator clean, protecting components and therefore extending the life of the concentrator. Some of the approaches suggested here, and expanded on below, could be retrofitted to an existing oxygen concentrator system.
Pre-filtering Air
The technology used to pre-filter air before it enters an oxygen concentrator needed to be low maintenance, passive and relatively low-cost. Two main approaches were identified: firstly use of a cyclone separator and secondly using a larger area passive filter that is more effective than existing gross particle cabinet filters. The solution needed to not restrict the airflow more than the existing gross particle filter.
Exploring the use of a Cyclone
Cyclones are passive filtration devices. Air is pulled in through the chamber of the cyclone, and the shape of the intake causes it to move in a cyclonic motion. A centrifugal force is created by the fast circular air flow, forcing the heavier dust particles into the wall of the cyclone chamber. As they hit the wall, they lose their velocity and fall down into a container below the cyclone, where the dust can be collected. The efficiency of a cyclone at removing different particle sizes is directly related to its geometry. From research a set of standard dimensions have been defined. Typically all dimensions of the cyclones are related to its diameter, the other dimensions are worked out in proportion according to a standard geometry, according to the desired particle size to be captured. Optimising the design for a particular application often requires detailed CFD (Computational Fluid Dynamics) simulation work and prototyping. An important factor in the design of a cyclone for an oxygen concentrator is the pressure drop across the system — this needs to be minimised. Cyclones pre-filtering dramatically reduces the amount of dust collected by panel filters and they require very little maintenance.
Smaller cyclones are generally better at removing smaller particles, due to the higher velocities created, but a single small cyclone alone would lead to higher pressure drop in the system. In order to achieve a low pressure drop, it was found that multiple smaller cyclones can be used in parallel, improving flow efficiency. In order to assess this, a sample of cyclones of varying size were 3D printed (see Figure 6a) with diameters of 25.8mm, 40mm and 51.6mm. This allowed tests to be conducted to understand which size of cyclone would be most beneficial and how the assembly could impact the performance of the device.
In order to get an idea of the effectiveness of different set ups, some simple tests were carried out. A larger cyclone was attached in series with the two smaller cyclones (as show in Figure 6a), the outflow of these cyclones was combined and flowed through panel filter (equivalent to the intake bacteria filter in a Devilbris Oxygen Concentrator). In order to get an indication of how well the different size cyclones performed an experiment was carried out: 100 grams of wood ash, 100 grams of sand, and 30 grams of talcum powder were measured our and mixed together. Suction was applied after the panel filter, creating a total air flow through the system approximately equal to the air required to run a 5L/min oxygen concentrator. The ‘dust’ mixture was then fed through each cyclone set up. The dust collected by each of the cyclones and the panel filter was weighed. This rough experiment that gave an indication of how effective different cyclone configurations were at pre-filtering dust and dirt.
In these initial tests the 40mm cyclone removed over 90% of the dust and dirt before the air reached the main cabinet filter. This compares to around 40% dust removal by a typical gross particle filter.
The cyclone system was then integrated into the oxygen concentrator, providing pre-filtering of air into the cabinet, which is then used for cooling as well as for oxygen process air for concentrating oxygen. Figure 7a shows a schematic of the set up. Air is drawn into the concentrator via a fan, it is pulled first through the cyclone system and then through a large effective cabinet filter before finally being passed through a large internal bacteria filter. The first internal filter could be a washable electrostatic filter — these are passive filters, where electrostatic charge is generated by air flowing through the maze of static fibres. Airborne particles are then attracted to and held by the static charge, until this is released by washing. Combined with the cyclones, this is an affordable and effective way of cleaning the air before it enters an oxygen concentrator. This protects the fine air and bacteria filter, which is not washable, and therefore significantly prolongs the life of the concentrator.
We explored the use of existing ‘snorkel’ cyclones (these are often used in off-road vehicles). Figure 7 shows how this fits onto our own oxygen concentrator prototype. The whole concentrator was placed inside a dust chamber. Fans were used to hold talcum powder in suspension. Off the shelf cyclone snorkels were shown not to be effective for this application — this is because the air flows provided to an engine is significantly greater than the flows in oxygen concentrators and the velocities in the snorkel were not sufficient to remove the dust to an acceptable level. These would need to be redesigned (modifying the geometry of the cyclone), in order to be effective; a similar test in a dust chamber was carried out with the multiple cyclone approach (described above) and this was effective, in repeated tests, with over 90% of the dust removed by the cyclone system, with a pressure drop of around 0.35 psi. Further optimisation is possible, using CFD, to increase the pre-filter cyclone particle capture and reduce the pressure drop, along with testing in a dust chamber in accordance with dust testing standards.
Larger internal dust filter
After passing through the pre-filtration the air was then passed through a washable pleated panel filter — this replaces the cabinet filter. The filter chosen offers more effective removal of dust than what is currently used in most oxygen concentrators, furthermore it filters all of the air entering the system. In order to achieve a low pressure drop, a large surface area filter was used (see Figure 8a), reducing the resistance across the filters. A G4 washable panel filter was selected, which is designed to capture pollen fog and coarse dust particles (>10um). It also captures any leaves, insects, textile fibres, ash, hair, and sand that comes through. The large area also reduces clogging frequency.
A washable filter was chosen as this is better for the long term serviceability of the concentrator. This combination of pre-filtration of all the air entering the cabinet, followed by a larger and more effective cabinet filter provides good protection of all the internal components against dust. When the filter does require replacing, this is easy to do (see Figure 8b). After this filter, the air then passed through a final intake bacteria filter before entering into the oxygen concentrator.
Other dust related considerations
If all the air going into the concentrator is filtered, it is then essential that the casing is also IP sealed in relation to dust — ideally to an IP dust rating of 6. In our design we also additionally IP protected the electronics (the PCB and power regulator) within the main enclosure. The box is made of aluminium and designed to dissipate heat. In addition to cooling the compressor, the air from the cooling fan is passed over this electronic protection box, to ensure everything stays cool. One other important aspect that we considered was the use of monitoring e.g. an alarm/phone notifications to alert maintenance personnel of the need for filters to be cleaned. This can be easily measured by monitoring the pressure drop across filters — relatively low cost pressure differential sensors are now available that make this kind of sensing affordable to integrate into any oxygen concentrator system.
Design of a self-cleaning vortex-cyclone type system to pre-clean air
Axial cyclone separators are used on some combustion engine applications [6] as a way to separate dust. The device is potentially valuable for pre-filtering in oxygen concentrators for low resource settings as it is self cleaning and known to be effective. In axial cyclone separators a helical swirl is used to impart swirling motion of the dust laden flow, dust moves to the outside of the flow of the tube and is separated (see Figure 9).
When a particle enters the swirl generator, it gets thrown radially outwards as the axial helix imparts a radial acceleration on particles in the air. It is carried to the outside of the helix and gets separated at the bottom, while clean air passes through to the oxygen concentrator. The main parameters affecting its separation performance (dust cut off size) and the pressure drop across the system are the filter diameter, blade length, number of blades in the helix and the blade angle in the helix. When designed correctly, for the expected inlet velocity, these systems can remove close to 95% of the dust.
We carried out some initial prototyping of this approach, and showed our current design could remove 88% +/- 6% (see Figure 10 and Table 2 below), but further optimization is possible, tuning specified parameters in order to obtain the greatest filtration efficiency with the lowest pressure drop. The approach is promising because it is simple and requires no maintenance. It does not replace the cabinet filter but ensures that it will last for significantly longer before requiring replacement. However, this approach only works when there is a fan drawing air into the concentrator, so it would be more difficult to ‘add on’ to existing concentrators without adding an additional fan. However, it is an interesting approach to be considered for next generation oxygen concentrator designs, and something we will continue to explore in our work.
How does this apply to protect existing concentrators
One of the easiest steps to protect existing designs is to add additional filtration to the outside of the casing to provide additional filtration to the process air — this will reduce the likelihood of blockage of the finer compressor filter. The filter should be washable and have a large surface area in order to add minimal pressure drop as you don’t want to risk increasing the running temperature of the compressor. This mitigates a number of the key failure modes relating to dust. One of the challenges of retrofitting though, is different concentrators have different designs of air intake and it is unlikely that ‘one size fits all’. Furthermore, it does not however mitigate dust getting into the cabinet, as this occurs mainly through the gaps in the casing — see for example Figure 3 and 4 above. Adding additional filtration here is likely to cause a pressure drop and therefore could potentially affect the cooling of the compressor, as the system was not designed with this additional filtration in mind. Another approach is to place the complete oxygen concentrator in another box, with a fan and filter on the outside, cleaning the air before it even reaches the concentrator — this is a relatively simple solution — of course there are other considerations to be made here, such as the exhaust of air and the cooling of the system, but this type of solution could be implemented without significant change to the concentrator and mitigates the need for a complete system redesign.
Conclusions
Dust is a major factor that contributes to issues with oxygen concentrators in many low resource settings. This is largely because devices are not designed to cope with the level of exposure to dust that can be experienced in some of these settings. Although there are some aspects can be solved by better maintenance schedules, many dust issues can be easily mitigated in future concentrators; some approaches have been suggested that could be implemented to achieve this and which we have been testing as part of a new concentrator design we have been working on.
We would be keen to get your feedback and hear your ideas on how to increase the resilience of oxygen concentrators to dust, so please do get in touch.
In the next blog we will explore resilience to Moisture.
References
- Oxygen concentrators offer cost savings for developing countries. A study based on Papua New Guinea, Anaesthesia, 46(3):217–9, Michael Dobson.
- Oxygen concentrators: A practical guide for clinicians and technicians in developing countries, Annals of Tropical Paediatrics International Child Health 30(2):87–101, 2010, Trevor Duke et al.
- Glob Health Sci Pract. 2013 Aug; 1(2): 249–261. Published online 2013 Aug 14. doi: 10.9745/GHSP-D-13–00037 PMCID: PMC4168575PMID: 25276537
- Modern Energy Services for Health Facilities in Resoure-Contrained Settings https://apps.who.int/iris/bitstream/handle/10665/156847/9789241507646_eng.pdf
- https://www.openo2.org/concentrator-graveyard
- See for example, ‘Conceptual and Preliminary Design of a Hybrid Dust Filter for Helicopter Engines’, Antonio Flippone, Sept 2017, European Rotorcraft Forum, Italy.
