How to minimise regrowth risk in water towers and large reservoirs by mixing optimisation - Part I
Water towers are considered fairly simple systems, yet their mixing dynamics are not as simple as you would think. Drinking water companies want to reduce the risk of bacterial growth by minimising Hydraulic Residence Time (HRT). Together with drinking water company Pidpa and BIOMATH (Ghent University),we've looked through the walls of a water tower and it's worth sharing what we've learned.
This blog post discusses what we observed during the 3D modelling of an existing water tower, and how we transferred the 3D results into very understandable 2D plots that enabled better decision making. Operational improvement (changing design and operation) will be handled in a second blog post. Do not spend too much attention on the specificity of the case itself. In this blog we share insights that are valuable for water and wastewater treatment in general.
Fig 1: A typical diurnal pattern of the water level (i.e. volume) in a water tower
A day in the life of a water tower
A water tower has a dynamic operation, matched to our daily lives. The water height varies throughout the day in predictable patterns. And yes, weekends and holiday periods obviously look different. The water tower we studied was filled during the night. Two major consumption periods can be distinguished: 1) waking up and morning activities, 2) coming home and evening activities.
P.S: wastewater treatment plants have similar but mirrored influent flow rate patterns.
Pidpa's mission is to deliver safe and high-quality drinking water. It's one of the large drinking water companies in Flanders (Belgium),operating in the Antwerp region. Water towers play an important storage and distribution role in some areas. Given their large volumes (100s to 3000m³), it is worth asking the following question:
What are the mixing dynamics inside a water tower throughout the day and week, and is there a way we can minimise the risk on ‘water aging’ by optimising the water tower’s operation and/or design?
The water tower we studied looks like this:
Fig 2: The 600m³ water tower under study, with the inner basin on the right. Not so many new towers are being built in Western Europe, but a lot are in operation. The way they were* designed varies enormously, just like the mixing.
(*) many of them are nice monuments
The questions we answered
A water tower is never emptied completely to safeguard the water supply. In this case, around 38% of the total volume remains at the end of a cycle. This is what we call the ‘older’ water. In this project, we wanted to answer the following questions:
- How are ‘older’ and ‘new’ (incoming) water mixed during filling, and what is their ratio in different zones of the water tower? Are there zones with high fractions of older water?
- What is the HRT (age) distribution of the water during long-term operation >24h?
- How can we improve the design or operation to minimise the bacterial regrowth risk?
Questions 1 and 2 will be answered now. Question 3 is the topic of our next blog post.
Why would you use 3D simulation?
3D simulation using computational fluid dynamics (CFD) is a very elegant - and the only feasible - way to answer these questions. Real measurements were practically not feasible and would have been very costly. Moreover, there's more to see - and thus learn - with simulations. CFD is an accurate tool for an application like this. Thanks to our in-house tools we could model several day cycles in a row, and we quantified the homogeneity in an understandable way, allowing for solid decision making.
Curious about the results? Here they are.
The impact of the filling location
A filling cycle starts with a non-submerged inlet pipe (Fig 3). The inlet is located somewhere near the wall, so it takes a while for the new water to ‘conquer’ the opposite side. We saw that filling from above the surface led to very inefficient mixing.
Key point #1: if you add water to water, they do not spontaneously mix. The diffusional force is really low and the incoming water creates an action that leads to an opposite buoyancy reaction of the older water mass.
Fig 3: New and older water mixing pattern during filling from above the water surface (see color scale for ratio new/older water; transparent = 100% older water)
Notice what happens once the inlet pipe gets submerged (Fig 4). Hydrodynamics change drastically with the new water penetrating the opposite side more easily and pushing the zone with older water to the middle (Fig 4, right).
Fig 4: New and older water mixing pattern during filling from below the water surface (see color scale for ratio new/older water; red ≥0.5)
The next blog post will contain a spectacular bonus video on those mixing patterns ;-).
Key point #2: how and where you locate your inlet and outlet impact your system dramatically.
Where the water goes during emptying
The outlet is located at the bottom, away from the center near the wall (Fig 2). During emptying, less energy is introduced and mixing intensity is less. The question is: where do the older water ‘packages’ go? Fig. 6 shows water transport during emptying. It becomes clear that not all zones are removed equally.
Fig 6: Where the water goes during emptying
Having understood how the system behaves, it is time to quantify the older water accumulation in different zones of the water tower during continuous operation.
Simplifying the CFD results to facilitate decision making
Looking at CFDs colourful drawings is one thing. The real value lies in:
- Processing the data as such that system performance is quantified and shown in an understandable way (in this case ‘how much older water is left and where?’)
- Optimising water tower operation and design using the accurate model at no risk and without the need for real-life measurement and experimentation (see next blog post)
At AM-Team we have developed our own methodology to take CFD results to the next level. The methodology we developed in collaboration with BIOMATH (UGent) facilitates decision making significantly. It is all about transforming the 3D results into very understandable 2D graphs. Our CTO Dr. Usman Rehman has put quite some PhD work into this.
First, we divided the basin into 24 compartments (3 pies with 8 pieces stacked on each other):
Fig 7: A compartmentalised water tower to quantify local older water accumulation
This allowed us to see:
- How much older water remains in each of the compartments
- How these values deviate from an ideally completely mixed system (CSTR)
Fig 8: Comparing reality with an ideal (theoretical) completely mixed system, which is often assumed
Form this graph it becomes clear that there is significant heterogeneity in both vertical and horizontal directions. The top region is nearly completely mixed, the bottom region contains significantly more new water than older water and the middle region contains most of the older water. The compartments opposite to the inlet/outlet region contain most of the older water, with M4, M5 and M6 (see also Fig 7) being the champions.
The impact after prolonged operation
After compartmentalization, we simulated 7 cycles in a row, originally starting from new water (0 days old). This allowed us to study the accumulation of older water throughout the cycles. After 6 days, the original water from day 1 seems to be removed. At first sight, the real water tower and theoretical CSTR do not differ that much. However, the black line indicates that a significant fraction of water with age 3 days or more always remains.
Fig 9: Water aging during prolonged operation. Bad mixing keeps significantly more older water and in a more concentrated form.
How you operate systems and define the critical design factors (eg, a slight change in inlet/outlet placement) will determine actual process performance and reliability. Real mixing behaviour and its impact can differ quite a lot from what is initially expected.
The next blog post will be the most interesting one: changing the operation and design to make this beautiful water tower a true champion.