The plant designers are unanimous that the precise estimation of desalinated water cost, on which the PPP project's success hinges, is extremely challenging. Uncertainty of the cost prediction inflates the consumer price of the water and the project risk.
The reason lies in poor prediction models of site-specific environmental data, construction, and operational costs. For instance, the construction cost contingency of 10-15% (covering inadequacy of prevailing cost estimating methods) adds 5-7% to the water cost. It is enough to lose the project.
Moreover, being built as XLS spreadsheets, the models cannot handle more than 3-5 possible combinations of parameters describing the project-plant life cycle economics. Such a coarse optimization is not enough to reach the ultimate minimum, which guarantees the project award.
Multi-variable optimization should be capable of comparing hundreds of scenarios, including the plant design and control philosophy modifications. Such an optimization is very similar to generative design. It is an AI-powered process where software creates and optimizes numerous scenarios by exploring solution spaces within constraints and goals set by the user.
To follow the generative design path, all the plant data shall be equally accessible and stored in the same database. Secondly, the plant design shall be paired with the plant digital twin (DT) simulating its operation and maintenance. Currently, crenger.com is the only technology meeting these fundamental requirements.
The annually averaged water cost is primarily defined by three inputs: the annual water production, the electricity consumption, and the capital cost. The first two values may be found by integrating the plant performance over the time of the year with a step not more than 2 hours to leverage the electricity tariffs. It means that the plant performance is calculated not less than 4350 times. The calculation takes into account seasonal variations in the seawater temperature, salinity, turbidity, and SDI.
Another reason for a high number of iterations is time-bound reversible performance degradation of the intake piping, cartridge filters, and reverse osmosis membranes. This procedure cannot be conducted manually.
Importantly, the integration result is automatically adjusted to the predicted levels of the plant unavailability. Its increase in 1% adds 0.5% to the water cost. Hence, the unavailability of 3-5%, often referenced in feasibility studies, is not acceptable in the PPP contracts. The current version of the plug-in component for the water cost prediction and optimization is bundled with the PlantDesigner software. It scans the plant design database and retrieves quantities defining the plant operating expenses listed below.
- RO membranes replacement
- O&M materials and parts
- O&M labor expenses
- Chemicals consumption
- Product remineralization
- Cartridge filters replacement
- Filtration media topping
These data and the prices for consumables and services are an input of a financial model (FM), which the above-mentioned component contains. An excerpt of the input/output for the 220 MLD plant is given below (20-year 100% loan, 6% IR, straight-line depreciation).
| Category | Value |
|---|---|
| annual production, 1000 m3 | 82786 |
| annual electricity consumption, MWh | 211327 |
| annual electricity expenses | US$20369 thou |
| specific electricity consumption, kWh/cu.m | 2.55269 |
| annualized electricity price, cent/kWh | 9.6 |
| plant capital cost | US$189243 thou |
| annual capital cost | US$16499 thou |
| RO membrane replacement | US$613 thou |
| O&M labor expenses | US$972.0 thou |
| remineralization expenses | US$332 thou |
| electricity expenses | US$20369 thou |
| O&M materials and parts | US$100 thou |
| chemicals consumption | US$2168 thou |
| cartridge filters replacement | US$194 thou |
| filtration media topping | US$3 thou |
| present value (PV) water cost, US$/cu.m | 0.498 |
Besides standard functions, the FM includes the water cost indexation (WCI) and risk and sensitivity analysis (RASA) tailored to desalination megaprojects.
WCI is intended to compensate for variations in operating costs following the market prices of energy, chemicals, reverse osmosis membranes, labor, and raw materials over 20-odd years of the plant service. It is a fundamental element of a sound risk allocation strategy.
WCI is part of the PPP contract describing the indexation procedure. Its core is an equation linking the contract (base) price to the current year one. Its basic form for consumables is given below.
C = Cbase + a1*(p1 – p1base) + a2*(p2 – p2base)+…, (A)
where a1, a2 – constants, p1, p1base, p2, p2base – base and current market prices
It is generally acceptable to link labor and spare parts – fixed O&M costs - to producer price indexes (PPI) published monthly by national bureaus of statistics.
C = Cbase + Ci1*(I1/I1base – 1) + Ci2*(I2/I2base – 1) + … (B)
where Ci1, Ci2 – the water cost components related to indexed expenditures, I1base, I2base – the PPI indexes for a base year, I1, I2 – the PPI indexes for a current year.
As follows from equation B, the water cost indexation is tied to the cost breakdown. It shall be disclosed in the PPP contract. Finally, the project price indexation clause shall be added to the contract to make possible indexation of annual capital cost. It is extensively discussed in "Project Price Indexation".
RASA is a way to jump from the predicted PV water cost to the minimum water price. Risk analysis quantifies the probability of planned water sales offset to unforeseen events like jellyfish inrush or algae blooms. Its principles are discussed in "Risk Prediction Framework" and "Modeling Disasters".
Sensitivity analysis identifies how much the final water price fluctuates based on changes in individual input variables. For instance, the contract may ask for decreased production in the wintertime. To assess the effect of the plant operation below its design capacity on the water cost, the straight-line load curve of DT shall be replaced with the customized one.