The chairs and machines are what the unit is judged on, and the water treatment plant is what decides whether it runs. How to analyse the source, size the train, build the loop and test it week after week.
A dialysis unit is bought as chairs and machines. It runs, or fails to run, on water. A haemodialysis patient's blood is separated from a very large volume of water across a membrane, several times a week, with none of the barriers that protect someone drinking the same water. Contaminants that are unremarkable in a municipal supply are a direct exposure here.
Setting up a dialysis unit as a whole, including chairs, layout and procurement, is covered separately on this site. This is about the plant room behind it.
Nobody can size a treatment train from a description of the supply. Get a laboratory analysis of the water that will actually feed the plant, taken from the point it will be drawn from, and take it in the season when the supply is at its worst. In much of Bangladesh that means the late dry season for concentration and the monsoon for turbidity and microbiology, which is an argument for sampling twice.
The panel should cover total hardness, iron, manganese, arsenic, fluoride, chloride, sulphate, nitrate, total dissolved solids, silica, total organic carbon, free and total chlorine, turbidity, pH, and a microbiological count. Chloramine specifically, not just free chlorine, because a supply dosed with chloramine will pass a free chlorine test and still destroy a membrane and haemolyse a patient.
Two Bangladeshi patterns come up repeatedly. Deep tube well supplies carry iron and manganese, sometimes arsenic depending on the district, and are usually hard. Municipal supplies carry a disinfectant residual and seasonal turbidity, and in coastal districts they carry salinity that varies through the year and pushes the RO harder than the design case. Both are manageable. Neither is manageable if the plant was quoted from a catalogue without seeing the numbers.
Stages are omitted from quotations to win on price, and the omitted stage is always the one that protects the next stage. Read a quotation by asking what each component is protecting.
| Stage | What it removes | What it protects |
|---|---|---|
| Break tank and booster pump | Nothing, it stabilises supply | The whole plant against pressure loss and back siphonage |
| Multimedia or sand filter | Suspended solids and turbidity | Everything downstream from blinding |
| Iron and manganese removal | Dissolved iron and manganese, by oxidation and filtration | The softener resin and the RO membrane |
| Softener | Calcium and magnesium hardness | The RO membrane against scaling, which is the usual cause of early membrane death |
| Carbon adsorption, two vessels in series | Chlorine and chloramine, plus organics | The membrane, and directly the patient |
| Cartridge filter | Carbon fines and remaining particulate | The RO pump and membrane |
| Reverse osmosis | The great majority of dissolved ions, bacteria and endotoxin | The patient |
| Distribution loop | Nothing, it delivers | Product quality, if it is designed to recirculate |
Three details on that table are where plants go wrong.
Two carbon vessels in series, with a sample port between them. The first vessel does the work and the second is the safety margin. Testing at the port between them tells you the first is exhausted while the second is still protecting the patient. A single carbon vessel gives you no warning at all, and it is the most common cost saving in a cheap quotation.
The softener needs salt, and someone has to buy it. A softener that stops regenerating because nobody ordered salt is silently passing hardness to the membrane. Put salt on the same standing order as the acid and the test reagents.
Single pass or double pass RO. Double pass gives a large margin on rejection and is worth it where the feed is high in dissolved solids, which includes most coastal district supplies. It costs more in energy and in reject water. Decide this from the analysis, not from the brochure.
Water leaving the RO is clean. Water arriving at the last machine on a badly designed loop may not be. Bacteria colonise pipework and form biofilm, and biofilm is far easier to prevent than to remove.
Storage tanks are a judgement call. A tank gives buffer capacity for peak demand and it is also the largest single opportunity for microbial growth. If you fit one, it needs to be closed, vented through a hydrophobic filter, sloped to drain fully, and included in the disinfection routine.
Product water demand is not the sum of the machine flow rates. Build it up as follows, and get every figure from your own machines and your own schedule rather than from a rule of thumb.
Then check the supporting services. Feed pressure and flow at the worst time of day. Drain capacity for reject and for disinfection dumps. Electrical supply for the pumps, and specifically what happens to the loop pump during a generator transfer, because a loop that stops every evening is a loop that will grow biofilm. Space and access for changing membranes and media, which needs more room than the footprint suggests.
The ISO 23500 series sets out the framework for preparing and managing fluids for haemodialysis, covering the treatment equipment, the water, the concentrates and the dialysis fluid. What the unit needs is that framework turned into a chart on the plant room wall with a signature column.
Every one of those tests needs a reagent or a kit that has to be in stock. Order them with the plant and put them on the standing order, because a unit that has run out of chloramine test strips is a unit running without its most important daily check.
Ask for a documented sequence: system flush, disinfection of the plant and the full loop, rinse, residual test, then water quality results from the plant and from the last station meeting the standard, chemical and microbiological, before any patient is connected. Add a demonstration of the alarms, a written operating procedure in the plant room, and training for the technician who will actually run it, with a second person trained as cover.
Get the as built drawing of the loop with every branch marked, and keep it. When a station is added or removed in three years, that drawing is what stops a dead leg being created.
On the treatment floor, the chairs are the other half of the unit. Digiterm's Comfort range is built around long treatment sessions, with a 200 kg maximum patient load across the range and options that matter operationally: integrated weighing to 100 g accuracy on the Scale models, battery backed actuators on the Comfort-4 Battery so the chair can still be moved out of position during a mains failure, and blood resistant upholstery. In a country where an evening power interruption during a session is a normal event rather than an emergency, a chair that cannot be reclined without mains power is a genuine problem.