There is a widespread confusion between the desalination plant availability and reliability terms. The latter relates to the procured equipment failure rate and is the negotiation point of the project package whereas the first is not as it depends on the repair capabilities of the plant personnel and the adopted preventive maintenance plan (PMP). It necessarily includes such time-consuming procedures as CIP cleaning, the membranes reshuffling and replacement and the seawater intake pipe pigging.
Quantitatively availability is always less than reliability. Depending on the plant design the latter varies within 97.5 – 99.5%. Availability below 95% is not rare. So only the plant reliability defined as
(1 – repair_time*capacity_drop/8760)*100%
is subject to the performance guarantee.
When a plant is reliable, maintenance costs decrease, unplanned outages and incidents are few and far between, the cost of production goes down, and profits go up.
The plant reliability design requires comprehensive system approach and is a compound result of the following vectors.

  1. Equipment design reliability,
  2. Spinning reserve capacities,
  3. Backup capacities,
  4. Plant structural reliability.

Requirements for more reliable design and construction are set out in the equipment specification. In practice the equipment is considered reliable if it has enough similar application references, good operating records and proven sufficient operation life.
The pump will be more reliable if it complies with the following criteria

  1. low level of vibrations,
  2. operation in the vicinity of the BEP (best efficiency point),
  3. sufficient NPSH (net positive suction head),
  4. low specific speed and suction speed,
  5. bearing temperature below 75oC,
  6. correct installation of piping (without load transfer),
  7. proper runs of piping to prevent hydraulically induced vibration,
  8. provision for temporary strainers,
  9. special design of volute decreasing radial forces on the shaft bearings,
  10. usage of the impeller diameters below maximum values,
  11. application of the cartridge mechanical seals,
  12. check for the design critical speeds,
  13. application of corrosion-resistant materials

For optimally selected and maintained pumps MTBF (mean time between failures) is far above 5 years, average values being 2.5 – 4 years and pump repair costs being $10,000 - $25,000 (the author estimate). This does not include lost opportunity costs.
The pump motors bigger than 1000 kW shall be ordered with reduced vibration levels and shall be designed at least for 300 starts per year. The motor life may be increased substantially if ‘soft’ start is used.
Interconnecting piping is rarely a source of trouble if the following points are taken into account.

  • sufficient difference between process pressure and the design limits,
  • thermal expansion analysis,
  • keeping velocities within the range applicable to the selected construction material,
  • proper mechanical support damping hydraulically induced vibration,
  • usage of weld-neck flanges for ANSI class #300 and above,
  • application of corrosion-resistant materials

Valves last longer if their sizing and specification are done with the following points in mind.

  • proper safety factor for actuator size,
  • reliable control without valve oscillation (due to the valve over-sizing),
  • application of corrosion-resistant materials.

The membrane manufactures guarantee mechanical integrity of the reverse osmosis membranes if and only if the pressure feed variation is kept below 1 Bar/sec during the plant startup and stop. At present practically no seawater desalination plant complies with this requirement – start/stop procedure is not controlled by interlock on excessive pressure variation. For new plants with the 16-inch membranes, planned to stop daily this limitation on the pressure feed variation may turn into a bottleneck problem hard to crack.
Spinning reserve capacity is the difference between the required design capacity and the rated one. Centrifugal pumps working at BEP always have rotating reserve capacity over 10%. Drives - AC motors and variable speed drives - are sized by 10-15% over the maximum continuous load. At low ambient temperatures motor may be overloaded by 5 – 10%. Dosing pumps are always selected with high redundant capacity sometimes surpassing 100%.
If the pumping station is equipped with 5 pumps each having 20% spinning reserve capacity and driven by VSD, than the capacity drop caused by one pump failure may be recovered to 80% by overloading the rest of the pumps. So there is no need in backup capacity. No tender addresses this issue.
Backup capacity is an effective means of providing high reliability for equipment modules connected in parallel. For instance, at the normal operation conditions all dual-media filters are in operation. In case one filter is shut down, its load is re-distributed between the balance of the filters without impairing the product quality. For seawater desalination plants working at 65 - 75 Barg, the pump backup capacities are very expensive. To reduce capital costs and given the axially split pumps high reliability, instead of the installed pump only it's rotor with impellers, bearings and mechanical seals may be kept in the warehouse. Another point to consider is that in warm surface seawater (above 28oC) even superduplex steels are subject to pitting and crevice corrosion (but at substantially less rates than the SS316 steel).
Structural reliability analysis (SRA) tries to assess the whole system reliability if the reliability data for the system each equipment piece is known. This analysis is an important part of the plant conceptual design starting from comparison of the plant different process schemes and the equipment unit sizes and quantities. Such comparison objective is to minimize the life cycle costs.
To the best of my knowledge SRA is never done by the book primarily due to a lack of the reliability database, software tools, and special training.
Lightweight version of SRA treats the desalination plant as a chain of processes – seawater pumping, pretreatment, feeding to desalination units, SWRO membrane arrays, energy recovery and , finally, post-treatment. The plant is generally admitted reliable if any of the above-mentioned chain pieces has the same level of reliability.
The reference point for SRA is the pressure center design – the term coined by the designers for the Ashkelon desalination plant (Israel, 2005). (Power engineering uses the cross-linked design term. One cannot help noticing that the desalination industry exactly follows the steps of the power engineering development of seventies – transition from the cross-linked design to the packaged one.)
The pressure center offers highest structural reliability and energy efficiency. On the other hand it has the following disadvantages (in the order of impact - lessons learnt from the power engineering history).

  1. long construction time,
  2. low level of standardization,
  3. high costs of civil engineering,
  4. complexity of operation,
  5. low level of maneuverability,
  6. high costs of interconnecting piping.

Due to the above-mentioned points the pressure center design is not a preferred choice for the majority of the modern desalination plants.
The reliability of the desalinated water production shall be treated differently from the reliability of the product chemistry maintenance: the product consumers are always tolerable to the temporary offsets from the product time-averaged quality criteria.
For example, if the seawater desalination plant produces technical-quality water for the nearby iron ore refining factory, the only meaningful requirement would be the water non-corrosivity corresponding to the LSI index above -1. Non-treated water (at LSI below -1.5) may trigger corrosion of the construction steels but at the rates substantially lower than those observed in seawater for SS316 and the duplex steels - main construction materials of the desalination plants. Under the circumstances, the post-treatment system with the ‘quality’ reliability of 90% would be an acceptable solution. In the case of the lime milk dosing system this figure may mean 30 – 50% (!) decrease in the installation costs as compared to the system with the 98% reliability.
Structural reliability requirements shall be relaxed for the equipment or systems with the average annual load substantially less than the installed capacity.
Let us consider the seawater desalination plant with Boron removal system (BRS) treating only part of the permeate quantity produced in the first pass (split RO scheme). Due to strong effect of the seawater temperature on the permeate quality, its flow rate to BRS varies from zero at winter time to 35-40% in summer. Such operation reduces the BRS installed capacity annual utilization to approximately 4500 hours. For such units the design criteria shall foster the design ruggedness and inexpensiveness even at the expense of power efficiency and shall allow for the temporary offsets in the product quality.
By the same token, auxiliary systems like CIP and chlorination (for intake and posttreatment systems) with the installed capacity usage less than 100 hours annually shall not contain any reserved capacities.

Table 1. Excerpt from the pump reliability database
1 Operation continuous
2 Power, kW 1000
3 Application high pressure booster
4 Equipment class axially split centrifugal pump
5 Manufacturer model [*]
6 qty 8
7 Cumulative operation time, hours* 140000
8 Down time, hours 104
9 Unavailability 0.000742857
10 MTTR, hours 8
11 MTBF, hours 21875
12 SM downtime, hours 98
13 SM unavailability 0.0007
14 MTTM, hours 6.5
15 MTBSM, hours 9285.7

Where MTTR - mean time to restore, SM - scheduled maintenance, MTTM - mean time to maintain, MTBSM - mean time between scheduled maintenance instances.

Plant reliability report sample