Designing Oxygen Concentrators for Low Resource Settings: Part 2 Resilience to Moisture
This is part 2 of a 3 part series on some learning from designing oxygen concentrators for low resource settings.
Part 1: Designing Oxygen Concentrators for Low Resource Settings: Part 1 Resilience to Dust
Introduction
Zeolite, which is commonly used as a sorbent material for oxygen concentrators, has a strong affinity for water. Moisture is strongly adsorbed onto the surface and is difficult to remove, greatly impacting on the performance of oxygen separation — both impacting the capacity as well as the selectivity; only a small amount of moisture (~1% by weight) [1, 2], prevents nitrogen from being adsorbed. Consequently, if zeolite is left exposed to the air, moisture is adsorbed and the material is ‘poisoned’.
One particular challenge related to moisture penetration in zeolite beds raised in relation to ‘shelf life’, is the length of time that a new unopened concentrator can be stored, before you need to be concerned about whether it is fit for use. UNICEF have reported stories of large numbers of oxygen concentrators being warehoused for over a year and then being unfit for use when finally required for an emergency. This is a significant issue, because sometimes oxygen concentrators are stored on wards for significant periods of time, with hospitals relying on them to perform when they are actually needed.
One of the reasons this challenge is exacerbated in many low resource settings is that often supply chains of spare parts are poor and replacements are difficult or time consuming to procure; this can lead to significant oxygen concentrator down time as shown in the study by Beverly Bradley et al. [3], where a faulty sieve bed would require removing the machine from service and might lead to a downtime of over 100 days. Ways to improve the resilience to moisture and ways to bring concentrators back online more quickly, are valuable for maintaining oxygen supply in low resource settings.
Why do oxygen concentrators have a limited shelf life?
Pressure swing adsorption (PSA) works by using a material known as an adsorbent, which preferentially attracts nitrogen molecules compared to oxygen. Adsorption is the accumulation of molecules at the surface of the zeolite. This occurs because there is an attraction between adsorbate (e.g. nitrogen and oxygen) molecules and the porous adsorbent surface. Adsorbents are crystalline porous structures, with a large surface area per unit mass. At a given pressure and temperature, different molecules interact with the surface differently and some will adsorb preferentially onto the micro porous surface. The challenge is that typical zeolites used in oxygen concentrators preferentially adsorb polar molecules, such as water, which has a strong polar van der Waals force and therefore preferentially attaches to the surface of the zeolite and reduces its capacity to adsorb nitrogen; zeolite has very high affinity for holding water and carbon dioxide molecules. These molecules block pores for nitrogen removal. In order for an oxygen concentrator to function correctly, the molecular sieve should have a water content ideally below 1%, whereas a concentration of 2% is reported to severely affect performance and 5% will render the system unusable [4].
There are two main types of adsorption, physical adsorption and chemical adsorption. Physical absorption can be more than a single layer and is easily reversible. On the other hand, chemical adsorption involves the formation of bonds between the adsorbate and adsorbent surface in a single layer. When water comes into contact with zeolite it can form a physical electrochemical bond and effectively blocks the pores, and reduces the adsorption capacity of the zeolite material. When the bond is ‘chemical’ adsorption, it is hard to regenerate — it is not easily reversible. In normal operation water is cyclically adsorbed onto the surface and then desorbed, but over time the interaction of CO2 and water with a zeolite can lead to irreversible chemical adsorption throughout the bed, after which the zeolite cannot be regenerated. This contamination is the cause of the gradual long term degradation of sieve beds.
When in operation and a concentrator is in cycling, there is a small purge between the columns generating a very low dew point gas, which passes through the regenerating column, and facilitates the desorption of nitrogen and moisture from the zeolite. This is what enables effective regeneration; regularly running the concentrator therefore significantly prolongs sieve bed life. However, when a concentrator is not used for significant periods of time, moisture can gradually enter the system by diffusion and contaminate the sieve beds. Without the regular purging cycle they become spoiled beyond regeneration by cycling — but how does moisture get into a sieve bed?
The Challenge of Reduced Shelf Life
Exposure of the sieve bed is one of the primary causes of reduced shelf life. This can occur, for example, with rotary valve systems that do not close off the sieve bed to the atmosphere when they are switched off, which is not uncommon practice in many existing off-the-shelf concentrators. It can also occur due to leaks in the system, and vapour diffusion through plastic components and the use of flexible silicone tubing — which has free volumes that permit gas diffusion.
Many issues, however, with moisture diffusing into the sieve bed and causing degradation occur while they are in use in clinical settings. This often occurs due to a problem that has been undiagnosed, such as a leak or faulty valve. For example, there were reported cases in Oxygen Colab meetings of one way valves failing, leading to moisture passing back into sieve beds from a humidifier, leading to irreversible degradation. Although these are not strictly shelf life issues in the purest sense, they come back to the same root cause — of moisture and CO2 degrading the sieve bed and rendering it unusable.
Colibri carried out an interesting analysis trying to estimate the mean time to failure of sieve beds when a concentrator is in ‘off-mode’; i.e. estimating the shelf life of a concentrator [5]. In this analysis they predict irreversible zeolite contamination within 4.6 months.
How and when are Oxygen Concentrators affected? The journey of an Oxygen Concentrator
An oxygen concentrator begins life typically under well managed environmental conditions where zeolite columns are rapidly packed in humidity and temperature controlled environments. At this stage, quality management, including, for example, leak testing the concentrators before they leave the factory, is an important factor in ensuring a long shelf life. Without strong quality management systems in place, the overall reliability of oxygen concentrators has been shown to suffer.
After being fabricated and tested, oxygen concentrators are either shipped or travel by air to their destination. Shipping containers, in which goods are transported, do not generally have controlled environments. For goods transported in containers, it is not uncommon for 10% of overall products to be damaged en route due to moisture related damage [6], this could be detrimental for oxygen concentrators. Concentrators can be in shipping containers at least 45 days, often over 60 days.
When concentrators are transported by air, travel times are of course much shorter. At 35,000ft temperatures outside an aircraft can fall as low as -50C, however, although cargo hold temperatures are not tightly controlled, they are generally kept above freezing (typically 5–15 C). It is important to bear these variations in temperatures in mind (as compared to the temperature at which the sieve beds are filled), when considering the likely air volume changes in the columns during transport due to the effects of temperature swing adsorption.
After transportation, concentrators are often then warehoused for a period of time before distribution. This might be at the supplier or procurement agency. UNICEF have reported that this can be between 1 week to 12 months, but typically will be around 2 months before units are used. Some warehouses will have environmental control, but it should be assumed that the majority will not have this in low resource settings. If the concentrators are not packaged in moisture resistant packaging or have sieve beds which are exposed to the atmosphere, sieve bed degradation will commence.
When concentrators arrive at their place of intended use, it is not uncommon for them to sit idle for periods of time in between patient use — between one week to many months, depending on whether the system is being used as a primary oxygen supply or as a backup system. Some can sit idle for considerable periods waiting for spare parts. The use case significantly affects how long it will sit in storage. It is during these idle periods when the zeolite deteriorates. Although user behaviour and usage of the concentrator is not in the strict definition of shelf life, it is an essential factor to consider in the design of a concentrator as intermittent use may lead to an oxygen system being unable to deliver when it is suddenly required, which can have potentially critical health impacts.
So how can the life of a sieve bed be maximised and are there ways to regenerate sieve beds to increase their useful life?
Improved packaging for Transport
One challenge that has been highlighted through the workshops we have been involved with is the need to protect an oxygen concentrator when in storage. This might be the complete unit, or it might be replacement sieve beds that are kept for when those in the concentrators need replacement. There are readily available packaging materials that are used in other industries to protect equipment from moisture which can be used to package systems to keep them safe in storage. For example, such materials are used in the pharmaceutical industry for the packing of pills, capsules and tablets, these include composite materials such as PET/PE coated with aluminium — there are various different composites available. These Aluminium blister foils offer full impermeability toward water vapour, gas and light. The PE or PET offers improved mechanical strength and prevents ruptures in the aluminium foil [7]. The performance of different technologies is expressed by its water vapour transmission rate (WVTR). A detailed assessment of the suitability of these materials is required. Technologies such as these are well developed and approved and available for use in the pharmaceutical industry — we propose that they could potentially be used for packing of oxygen concentrators and replacement sieve beds to increase shelf life.
Ensuring the sieve beds are isolated from the atmosphere when the system is idle
It is essential that there is not an open path from the environment to the sieve beds; they must be isolated from the atmosphere when not in use as well as when in storage, and especially if they are not being stored in a climate controlled space. Some oxygen concentrators, such as those designed by AirSep already implement this feature, with valves that close when not being powered.
If there is an open path between one of the sieve beds and the exhaust when the concentrator is not cycling [8], then a one way valve can simply be added to the exhaust of the system to ensure there is no backflow of air into the system (see Figure 1). However, it is essential that the valve chosen is both large enough and has a very minimal ‘cracking pressure’ (the minimum upstream pressure — pressure differential between inlet and outlet — to open the valve and allow flow through it). This valve does not need to ‘hold’ any pressure (e.g. compared to other valves in the system), it simply has to stop ambient air flowing into the concentrator.
On the Devilbiss 525KS, depending on the way the valve closes when the power goes off (e.g. if there is a sudden power failure), one of the sieve beds can be left exposed to the atmospheric air. Adding a check valve between the muffler and the rotary valve prevents the exposure of the beds to moisture. In our testing with a Devilbiss 525KS concentrator with this set up, the valve was shown to not affect the efficient operation of the appliance, but did it prevent the risk of moisture entering the sieve beds when the concentrator was not in operation.
Another pathway that can also potentially expose the beds to moisture is through the compressor: when the unit is not cycling some designs of concentrator leave a pathway open. When a concentrator is not used for a sustained period of time, moisture can gradually enter by this route and contaminate the sieve beds. Valves can be introduced in order to close off this pathway. For example, pneumatic valves are available that will open when driven by pressure and close (with a spring) when there is no flow (and the compressor is off), this type of system can be used to isolate the input. Another even simpler option, that we have found to be effective in lab testing, is the use of another one way valve, facing in the direction of flow. Because one way valves have a ‘cracking pressure’ — i.e. there needs to be a very small differential between either side of the valve for them to operate, at rest there is insufficient atmospheric pressure to open the one way valve. Thus it remains sealed to the atmosphere when the compressor is off and the concentrator is at rest. This of course does introduce a small increase in the pressure drop across the system, however, this is very small and appears to have little effect on the operation of the system, while reducing the diffusion of humidity into the sieve beds when the concentrator is idle. We propose that this would be particularly useful for oxygen concentrator devices that do not have another isolation valve (for when the system is off) and it would be easy to retrofit.
Regularly running the Oxygen Concentrator
During operation, when the concentrator is in cycling, there is a small purge from the column generating a very dry low dew point ‘light’ gas, which passes through the regenerating column and facilitates the desorption of nitrogen and moisture from the zeolite. This is what enables effective regeneration. Regularly running the concentrator therefore significantly prolongs sieve bed life. For example, it has been shown that simply running the concentrator for about 2 hours every 3 months keeps the degradation quite low and significantly prolongs the life of the sieve bed. The challenge is how to either automate this or better encourage good user behaviour. One approach that was suggested in a workshop was to design some sort of indicator system to feedback to users when the concentrator has not been run for an extended period. E.g. this might be a indicator light, or a text message to the user to encourage them to run the system for a period of time in order to maintain sieve bed health.
Ensuring there are no leaks
By monitoring some simple parameters over time (e.g. pressure in the columns) it is possible to detect if there is a leak in the system, with leaks being indicated by a considerable pressure drop. Digital monitoring of the device over time, therefore provides an approach to ensure that there are no leaks that would create a moisture pathway that could lead to the contamination of a sieve bed. Users (or service providers) could be sent notifications to give an indication that there is a potential leak in the system. This can then be fixed quickly and the of life of the sieve beds prolonged.
Selecting tube with a low water vapour permeability
Silicone rubber is universally regarded as one of the best-in-class elastomers in terms of flexibility, resistance to heat and durability. However, it has one of the highest vapour permeability rates among all types of rubber. The higher the vapour permeability, the higher the rate that moisture is able to pass through the material. Many oxygen concentrators use silicone rubber for some of their tubing and when a concentrator is left on the shelf for a considerable period of time, water vapour can permeate the system through the silicone tube, leading to contamination of the sieve beds. Many systems use PVC tubing, which is better but also has a permeability to moisture. The ideal material for the tube would have a high resistance to moisture permeability, for example using stainless steel pipes. However, this would increase the cost of the system by a significant amount and increase the complexity of manufacture (being less flexible). As shown in Figure 2 below, Teflon tubing (PTFE) has a low permeability to moisture and we explored its use in oxygen concentrators — however, its poor flexibility compared to PVC and silicone makes this more difficult, but with careful design (e.g. to avoid the possibility of kinking), it was possible. The choice of tubes is therefore a balance of permeability and flexibility. Figure 2 shows the permeability to moisture vapour for some commonly used tube material. Indeed, perhaps a hybrid option is possible — using a flexible inner tube, coated in a layer of material with a low permeability (for example, a silicone tube, coated with a moisture barrier coating). A detailed controlled trial is required to better understand the likelihood of sieve bed deterioration as a function of the tube material choice — there is currently a lack of data on this. However, from our understanding of material science, it is clear this will be an issue for machines sitting idle for long periods of time. It will of course depend on a number of factors such as length, thickness and surface area of the pipe within the oxygen concentrator system.
Making columns easily replaceable and perhaps refillable
One approach to solving the problem around humidity is not to make drastic changes to the design of the concentrator, but rather by making it very easy to replace sieve beds. This might be a plug and play type system, where pre-packed sieve beds can simply be clipped into place and removed by users, in a similar way that LPG canisters on gas stoves are removed and replaced on a regular basis. The whole system could be designed around the expectation that this will be required and even spare columns provided inside a device (in moisture protective packaging) on purchase, giving it an increased lifespan. Spent columns could be regenerated or refilled at a later date/returned to manufacturers. Column health could be monitored and reported to users through an indictor so that users are aware when column replacement is required. Advances in remote monitoring and IOT make this type of technology very feasible — potentially this type of system would work well in an ‘oxygen as a service’ business model.
Refilling and even reconditioning sieve beds in-country is also potentially valuable, rather than having to procure new ones which maybe hard to get hold of. Many manufacturers use sieve beds that were never designed to be refilled. OpenO2 provides some great guidance on replacing the zeolite in a sieve bed that is not designed to be opened [10]. However, if sieve beds could be designed to be refillable it would make replacement much simpler, quicker and possible with very limited tools. Appropriate zeolite can then be procured in bulk within a region (for example within a hospital or a service based organisation) and zeolite beds could be refilled and maintained as required. This is considerably less expensive than buying new beds. Again, this might work well within an ‘oxygen as a service’ business model.
As part of our investigations at CREATIVenergie we developed refillable sieve beds. They consist of bayonet style end caps (push and twist) which makes it possible to easily remove the columns, refill with fresh zeolite and replace. These could be manufactured in different sizes with different length columns for different makes and models of oxygen concentrator. Figure 3 below shows the design of the end cap with a column fitted onto a Devilbiss 525KS concentrator, making it much easier to replace the zeolite as and when this is required. A disk piston slots in and a spring retains the zeolite.
Reducing the moisture load on sieve beds
The challenge of moisture is not only related to the time when a concentrator sits idle. Depending on the environmental conditions (temperature and humidity), there might be between ~9g (e.g. in the UK on a relatively dry day) and ~50g of moisture in the air (e.g. in a very warm humid location). For oxygen generation in a PSA cycle, the dew point in the sieve bed needs to be around -55°C. For a 5 L/min concentrator, this amounts to between 0.5L to 5L of water per day that needs to be expelled from the system, depending on the humidity. For a concentrator operating in a very humid environment, there is much more moisture that passes through the sieve bed compared to a climate controlled clinical environment in a western hospital. Over time, this increased load on sieve beds leads to an accumulation of moisture and the beds will fail sooner than in a less humid environment.
Currently, in order to handle the removal of moisture, typically pressure swing adsorption concentrators use multi-layer beds, for example, using silicon gel, activated alumina or 13X before a more adsorbent lithium based zeolite. This is supported by the work of Rege et al. [11]. In their study they conclude that activated alumina (a-Al2O3) adsorbents appear to be the best for water removal and 13X zeolite is the best choice for carbon dioxide, also an important contaminant to remove. They suggest using a first layer of activated alumina, followed by 13X in a ratio of 7/3.
The greater the humidity of the environment, the longer the length of penetration of moisture into the sieve bed. Adding thicker pre-protection layers in the sieve bed, as described above, can be beneficial in these circumstances. However, when doing this we must bear in mind that using the multi-layered bed approach is going to decrease the energy efficiency of the concentrator. This is especially true when using zeolites such as activated alumina, as these take up volume but do not contribute to oxygen concentration. This can be achieved on existing concentrators by adding a small protecting ‘cartridge’ after the compressor in line with the sieve bed. We trialled this and showed it to be an effective approach to counteract, but not eliminate, the challenge of bed failure caused by moisture.
In order to get an indication of the energy impact of adding additional protection to sieve beds after the compressor an experiment was carried out. The oxygen concentrator flow was set to 0.5L per minute [12] and the BLDC (Brushless Direct Current) air compressor power was adjusted to use the minimal energy possible to generate just over 90% concentration. Protection cartridges were then added after the compressor and before each sieve bed, as shown in Figure 4. The power of the BLDC compressor was then adjusted in order to once again generate over 90% concentration of oxygen with the cartridges attached. The results of this are shown in Table 3 below. The volume of the cartridge was 0.34 litres. The experiment was run for an extended period of time in order to ensure that the results were representative and reproducible. This size of cartridge led to 112% increase in the energy consumption of the unit under the given flow rates. This is a significant increase in energy; this part of the system is not producing oxygen and therefore the compressor has to work extra hard to pressurise each bed. This gives an indication of the impact of adding this type of solution to an oxygen concentrator — of course using a smaller protection carriage would have less impact in terms of energy consumption, but it is a balance between achieving a good level of protection and the energy efficiency of the system.
How can the load on the bed be further reduced?
Another way to improve the lifetime of sieve beds operating in high humidity environments is to seek to further reduce the load on the sieve beds, effectively pre-treating the incoming process air to reduce the moisture to more tolerable levels. This can be done in several ways. In this section we are going to consider some of the options alongside their benefits and drawbacks.
One of the major aspects to bear in mind is that almost every approach taken to protect the sieve beds is going to consume energy: if pre-drying with absorptive material such as silica gel, this will need to be regenerated; if adding additional layers within a bed, or adding an additional removable zeolite cartridge after the compressor and before the sieve bed, more power will be required on every cycle; if a moisture membrane filtration approach is used, this will likely increase the pressure drop across the system. The choice will therefore be a balance of energy cost, practicality and maintainability. If a system is designed to be resilient in the most humid environment, it likely will have a lower energy performance in an environment where humidity is much lower — while potentially also costing more.
Reducing the dew point with passive cooling
One simple thing that can be done is to pass the compressed air through a copper tube, which is cooled by ambient air to encourage it to condense. This reduces the dew point of the ambient temperature of the air somewhat. The air can then be passed into a self draining coalescing filter to separate the water from the air stream. It does not remove enough moisture to reach anywhere near the dew points required for oxygen concentration, but it does have benefits, especially in very humid locations. Furthermore, it has no moving parts and is a well established approach that is used with compressed air in industrial applications. Although some oxygen concentrator designs do implement this approach, many do not (e.g. the Devlibiss 525 KS does not), and it is something which could be readily added to any existing oxygen concentrator to reduce the moisture load on the sieve beds.
Pre-drying incoming air with desiccant
This can be done by passing the air through a silica gel filled column before the air enters the compressor. This is something that has been done on a number of open source designs [e.g. 13]. Silica gel can absorb around 40% of its weight in water, so a kilogram of silica gel in a cartridge will capture up to a maximum of 0.4 kilogram of moisture — compared to the amount of moisture that passes through a 5L oxygen concentrator in a humid environment per day, this is a relatively small amount. Therefore a significant amount of silica gel would need to be used in the most humid environments. Even if the silica gel is only used to reduce the relative humidity by a small percentage, after a relatively small amount of time it will become exhausted. Therefore, unless regenerated this will only delay, rather than prevent, the failure of zeolite beds. There are also many desiccants available [14] and lessons could be learned from significant innovation around desiccants and desiccant based drying solutions in the compressed air industry, as well as innovations in dehumidification in buildings, when designing a pre-drying system for oxygen concentrators.
Although these solutions are relatively low cost, they often require regular maintenance, e.g. the simplest designs need regular regenerating of the desiccant. We know that in low resource settings there can often be challenges with regular maintenance of medical equipment. Based on this, it would be unwise to increase the amount of maintenance that is required for an oxygen concentrator by adding on pre-drying cartridges that must be regularly removed and regenerated. Furthermore, if the cartridges were not removed and regenerated, it could even exacerbate the problem, by continuously passing the incoming process air through saturated desiccant before the sieve beds.
Refrigeration drying
Small refrigeration drying systems are available and used in the compressed air industry (e.g. non-cycling refrigerant dryers), which could be used to remove moisture from the air for oxygen concentrators. They can be relatively low power (typically ~50–200W, depending on the size), but there would be a considerable capital investment required to include them in an oxygen concentrator system (currently circa an additional £1000) — a cost benefit analysis would need to be carried out for the particular scenario. In a first instance, these would likely be a module that would sit outside the oxygen concentrator unit and pre-dry the process air, but could potentially be integrated into a future concentrator design. However, given the cost is similar to that of an oxygen concentrator itself, its benefits would need to be carefully weighed up against other options.
Membrane drying
Moisture can be removed from the humid air by using a selective membrane, which allows the water vapour to pass through at high permeability resulting in pure water vapour in the permeate side. The membranes used in such separation processes are usually made of hydrophilic polymers with high vapour permeability and high selectivity towards air. The dehumidification membrane module can be integrated into the existing process, for example, the membrane modules can be placed after the air compressor to remove the water vapour before air is fed to the oxygen concentrator. These membranes are commercially available [15] . Such processes can be operated continuously and require limited maintenance. However, it should be understood that these typically require greater than 4 bar pressure to operate and continuously purge an amount of compressed air (~typically between 5–30%) and so in addition to the capital cost (~currently circa £700) there is a significant energy cost to using these systems (due to the purge and pressure drop across the system). There is scope for further innovation in moisture membranes for lower pressure applications like oxygen concentrators.
Select zeolite which is better at handling moisture
Reducing the dew point of air before the sieve bed is possible, but can add considerable cost and increase the energy demand of the system. Other ways to increase resilience is to improve the moisture handling capabilities of the zeolite material itself.
Different zeolites behave very differently to the presence of moisture and other gas contaminants in the system. Therefore, in designing a system for LRS, careful selection of the most appropriate zeolite is required. For example, LiLSX and AgLiLSX type zeolites are low-silica type zeolites, and have a strong affinity towards water vapour. Therefore, concentrators using these types of zeolites are more likely to suffer more quickly from a loss of capacity in tropical climates. In a study by Santos et al. [16] different zeolites were found to be affected by water vapour/carbon dioxide to different degrees — some zeolites were more damaged by the presence of water vapour than others under the same conditions. In particular they found that the adsorbent less affected by water vapour was the MS C 544 made by Grace Davison. The choice of zeolite used in a concentrator should therefore be carefully considered at production and procurement stages. Existing zeolites are optimised for energy efficiency and adsorption capacity (e.g. to reduce the weight of a concentrator) but there is scope and benefit for innovation related to developing adsorbents that are more resilient to moisture, allowing oxygen concentrators to better operate, and for longer, in high humidity environments.
Sieve bed regeneration
It would be useful, especially in humid climates in low resource settings, if there were approaches by which zeolite columns could be easily regenerated. The conventional way to regenerate sieve beds is to desorb the adsorbed water with a stream of hot gas at temperatures between 200–300 C, for a period of around 2–3 hours. Although this requires considerable energy, it is a process that is only done occasionally. Existing sieve beds are not designed to be regenerated, so manufacturers direct users to replace the sieve beds when they have deteriorated.
A device to regenerate beds used to be manufactured by Foothills Medical in Colorado. This was popular for a while until the cost of new beds came down, which made regeneration not worth it in the USA. They registered a patent in 1994 (Patent number 5,335,426) but it has now expired. It describes a method and the apparatus required for thermal regeneration of molecular sieve material used in oxygen concentrators. It provides a good amount of detail on an approach, which provides a good basis for anyone seeking to solve this problem. There is potential for a device of this type for low resource settings as it reduces the dependence on international supply chains and reduces the time an oxygen concentrator is out of service (regenerate on site rather than wait for replacements to be procured and delivered as well as removing the need to stock replacements which may deteriorate in storage).
Heat loss in the described process is high and a lot of equipment is required to carry out the process. We explored an approach which can be used and/or adapted to potentially enable either in situ regeneration or used to create a simple device that is able to regenerate sieve beds in-country. We tested it with silica gel desiccant, but it could be used for the regeneration of other types of desiccant and zeolite.
An electro-thermal sieve bed was created — containing a flat heater arranged in a spiral within the desiccant (see Figure 5). By turning on the heat, the adsorbent is heated directly, reducing system heat loss and simplifying system design, compared to existing ways of regenerating. Air can then be passed through the sieve bed (driven by the compressor), regenerating the system. Ideally, to achieve full regeneration, the air needs to be dry (e.g. the output from another working oxygen concentrator, or using a simple low cost desiccant dryer to dry the incoming air). We also trialled the system using air coming from the exhaust of the oxygen concentrator while it was running, and this was shown to be effective. A more detailed study on the degree of regeneration that is achievable with this approach would be valuable. The pre-drying column we prototyped reduced the dew point by over 50% and was then regenerated using the internal heating element along with the air ejected from the oxygen concentrator.
N.B. If this approach was being used within a pre-drying desiccant system, it is important to consider the desiccant carefully. Silica Gel is effective, but for regeneration, the advice is that it is heated to around 120 C for 1–2 hours. There are other desiccants which should be considered, that begin regenerating at lower temperatures (therefore using less energy) for example activated (bentonite) clay is a well known low cost and effective desiccant. It is natural, non toxic and environmentally friendly. There is much innovation going on in this space.
Potentially Disruptive Technologies
There are a number of emerging technologies, which might in the future be more appropriate for generating oxygen in remote health care centres and moisture in the way in which zeolite is. There are numerous approaches being considered, which are at different degrees of maturity. These include:
- Paramagnetic oxygen separation — This is where a strong magnetic field is used to do the gas separation. There is good physics that suggest this should work and some early experimental work by a group in Colorado has demonstrated this in practice. This would not be susceptible to moisture.
- Using algae to generate oxygen — It is possible to produce large volumes of oxygen from algae through photosynthesis, a process which can be run directly from sunlight or from LED light sources. The algae is grown in water and thus humidity does not affect this process like it does zeolite.
- Ceramic Oxygen Generator — The ceramic oxygen concentrator has been developed by NASA and other research partners. It is a solid state device with no moving parts that is durable and not affected by moisture. It uses a type of ‘wafer’ that when heated, pulls oxygen out of the air. A prototype has operated for a number of years at a military base. Ceramic membranes have been intensively studied in the past two decades, by researchers in Imperial College London, Nanjing Tech University, and Dalian Institute of Chemical Physics. As well as NASA, Nanjing Tech University has also developed a prototype of a ceramic membrane based oxygen concentrator. These are not affected by moisture.
Innovations such as these, would significantly change the design of oxygen concentrators, and would be game changing for health clinics in low resource settings, enabling reliable access to oxygen in places where this has so far been challenging.
Conclusions
Moisture is a significant challenge for oxygen concentrator because of the strong affinity of zeolite with water and improving the resilience to moisture of oxygen concentrators is of vital importance to ensure that these devices are able to function as intended for long periods of time. There is significant potential for innovation and R&D in this area developing new technologies that address this challenge.
We are keen to hear your reflections what we have share on how it might be possible to protect oxygen concentrators from moisture. If you are interested in exploring this further, whether it be discussing, critiquing, trialling or optimising some of the ideas presented within this article, or sharing some other new ideas, we’d love to hear from you.
This is the second part in a 3 part blog series, sharing some of our learnings in designing an oxygen concentrator that is more resilient in low resource settings.
References and Notes
- Effects of a Readily Adsorbed Trace Components (Water) in Bulk Separation PSA Process: The Case of Oxygen VSA, Ind. Eng. Chem Res 2001, 40, 2702, Wilson et al.
- Contamination of Zeolites Used in Oxygen Production by PSA: Effects of Water and Carbon Dioxide
- A retrospective analysis of oxygen concentrator maintenance needs and costs in a low-resource setting: experience from The Gambia, Health and Technology Beverly D. Bradley et al., 4, 319–328 (2015)
- Regeneration of Molecular Sieves: regenerating or reactivating of aluminosilicate molecular sieve patent US4319057A (expired) https://patents.google.com/patent/US4319057A/en
- Accelerated Failure of Medical Oxygen Concentrators in Resource-Constrained Settings, Dylan Owens et al., 2021, https://colibrioxygen.com/wp-content/uploads/2021/08/Colibri-Co.-Accelerated-Failure-of-Medical-Oxygen-Concentrators-in-Resource-Constrained-Settings-9-Aug-2021.pdf
- For a good overview of the issues around cargo damage in shipping read “The Essential Guide to Cargo Damage, Types, Reasons, Prevention and Handling https://shippingandfreightresource.com/wp-content/uploads/2017/10/The-Essential-Guide-to-Cargo-Damage.pdf
- Voigt’s Pharmaceutical Technology, Alfred Fahr, John Wilsey & Sons, 2018
- see Accelerated Failure of Medical Oxygen Concentrators in Resource-Constrained Settings, D. Owens et al.
- For more information see ‘The Permeability Characteristics of Silicone Rubber’, Haibing Zhang, also see Permeability Coefficients of common polymers to moisture ‘How Weld Hose Materials Affect Shielding Gas Quality’, Paul Bhadha, Welding Journal 78(7):35–40. N.B. the permeation of atmospheric moisture into hose walls is a cause of defects in welding, particularly when when welding aluminium. and ‘Permeation of Gases and Vapours through films and thins sheet — Part II’ M. Lowax, Rubber and Plastics Research Association of Great Britain, (http://www.faybutler.com/pdf_files/HowHoseMaterialsAffectGas3.pdf, https://imageserv5.team-logic.com/mediaLibrary/99/D116_20Haibing_20Zhang_20et_20al.pdf
- See https://www.openo2.org/innovations
- Air-prepurification by pressure swing adsorption using single/layered beds, Salil U. Rege, Ralph T. Yang, Kangyi Qian, Mark A. Buzanowski, Chemical Engineering Science, Volume 56, Issue 8, 2001 https://doi.org/10.1016/S0009-2509(00)00531-5
- N.B. generating 0.5 L/min was chosen to ensure there was sufficient power in the BLDC compressor available in order to generate oxygen at a sufficient concentration when the protection cartridges were added.
- see http://oxygenator.getechprojects.com/index.php/sample-page/setup/prototypes/
- For example see https://www.l-i.co.uk/interests/catalysts-adsorbents
- e.g. https://www.ultra-filter.com/compressed-air/membrane-dryer/
- Contamination of Zeolites Used in Oxygen Production by PSA: Effects of Water and Carbon Dioxide, J. C. Santos, F. D. Magalhães, and A. Mendes* https://pubs.acs.org/doi/abs/10.1021/ie800024c#
