W.H. Dudok van Heel and J.D. van der Toorn (1988)
A biological approach to water purification: II.
A practical application: The Delfinaario in Tampere, Finland

From: Aquatic Mammals 14(3): 92-106

Table of contents

Original design
System hydraulics
The biofilter
Foam fractionators
Water composition
Nutrient level in the Delfinaario water
Recent developments


In this paper, the water treatment system of the Delfinaario in Tampere, Finland is described. This system makes use of biological and physico-chemical processes for the purification of the water (an artificial seawater mix) without the use of chemical additives. The system has been in use now for 2½ years and has proven to be easy to manage and has shown a good performance. The levels of nutrients in the water, except the nitrate-level, are well below the advised levels for seawater aquarium management and bacterial counts are low.

Note: this paper has been slightly adapted based on new data received after publication. Where relevant, these changes have been marked. Some illustrations have been added. At present, the system at the Delfinaario is still operational in essentially the same setup as discussed here.

Back to Table of Contents


Traditionally the water of inland oceanaria has been purified with the aid of large sand filters and chlorination. This method has been described by among others Andersen (1983) and Dudok van Heel (1983). Although chlorination can create acceptable water conditions for marine mammals, both authors agreed that it is not the ideal solution.

In the early days, before 1972, energy was cheap and obtaining fresh water and releasing wastewater presented no real problems. Therefore, there was no incentive for looking into alternative water treatment techniques. Why look for another method if there is a cheap method that gives acceptable results? However the situation changed: energy and chemicals became more expensive and an increased concern for environmental pollution caused an increase in the taxing of water and of waste-water dumping.

It marked the end of easy going and cheap management. A switch to new paths became essential. It became clear that there was not much room for improvement with the chemical water treatment. When chlorination is applied, regular replacements of large amounts of water are necessary to prevent the buildup of potentially hazardous chlorinated hydrocarbons (Duursma and Parsi (1976), Dudok van Heel (1983)) and of total organic carbon (TOC) (Spotte and Adams (1979)).

An alternative way of cleaning the water in oceanaria could be biological water treatment.

Water purification strategies based on biological processes have been used for instance in aquaria (Spotte (1979 a,b), Kinne (1976), Wolff (1981)) and in large wastewater purification plants (McKinney (1962), Fair et al (1968), Tri (1975), Conway and Ross (1980), Mudrack and Kunst (1986)).

There is no reason to assume that these processes cannot be applied to dolphinarium water purification. The processes involved have been outlined in part I of this paper (Van der Toorn, 1987).

It is the opinion of the authors, that biological water treatment is an alternative that creates a much healthier environment for the animals and in addition is more economical to run.

Back to Table of Contents

Original design

The Delfinaario at Särkänniemi in Tampere, Finland, officially opened on May 11, 1985.

Early in the designing stage, the decision was made to incorporate a biological water treatment system instead of the traditional chlorinated system. This decision had some consequences with respect to the operation, because in fact a biological filter is a living entity, which needs food and oxygen. In this system oxygen would be provided by incorporating a trickling filter. In such a filter the water is divided over a large area and cascades down, thus collecting oxygen. The food is provided by the excrements of the animals, in this case five subadult bottlenose dolphins, Tursiops truncatus. The greater part of the material excreted by the dolphins is biodegradable and will be taken care of by the micro-organisms in the biological filter. Because some materials are not converted easily or not at all, parallel to the biofilter a number of foam fractionators were installed for the removal of those materials.

Since foam fractionators will also remove organic matter, this setup creates a kind of buffer for the biological filter : if there is a sudden peak in excretion by the dolphins, the biofilter doesn't have to deal with all of it. The original design called for a 50-50 distribution of the flow between the biofilter and the foam fractionators. Since micro-organisms multiply rapidly when nutrients are available and also die rapidly if there is not enough, one can expect a fluctuation in the number of micro-organisms and thus in the purification capability of the system, when nutrient supplies are not constant. There will be a time lag between an increase in nutrient levels and a build-up of micro-organisms.

The division of the water as designed will lessen the fluctuation of nutrient levels and thus create a more stable environment for the biofilter.

The design also included a provision for recirculation of the sludge that is always formed by a trickling filter to feed the filter in periods of lower nutrient supplies from the pools. This would maintain the biofilter at a constant high level of activity. Since it was not possible to dimension this provision properly without more details about the system's performance it was decided not to install this right away. As it turned out, it was not necessary in this system but it is something that must be kept in mind when designing such a filter. It could improve the efficiency of a biological filter.

 Click here to load design drawing in current window or click here to load design drawing in a new window.

Back to Table of Contents

System hydraulics

The Delfinaario in Tampere has 3 pools connected by a channel. In the original design the show pool, with a volume of approx. 700 m3, would be served by 3 pumps with a capacity of 180 m3/hr each. Two of them would lead their water to the bio-filter, the other one would serve 3 foam fractionators, with a capacity of 60 m3/hr each. The holding pool, with a volume of approx. 260 m3, would be served by 1 pump of 180 m3/hr, which would lead all its water over 3 foam fractionators. The medical pool, volume 30 m3, and the channel, volume 24 m3, would each be served by one pump of 30 m3/hr, which would together serve one fractionator. All the water from the fractionators and the bio-filter would be collected in a mixing channel, where it would be forced through a limestone bed or a bed of broken shells, and would then be led to the pools by gravity.

Click here to load the pool layout drawing or click here to load this drawing in a new window.

The medical pool could be used as a quarantine, by leading the water from the fractionator it served directly back to the pool, bypassing the mixing channel. The water from the channel pump would then temporarily be led over the bio-filter. The total circulation would be 780 m3/hr. Since the total water volume, including the filter system is approx. 1250 m3. the turnover would be about 1 hour 40 minutes. Of the total circulation 50 % of the water would be led over the bio-filter and 50 % over the foam fractionators. The water returning to the pools would be treated with Photozone (photochemically generated ozone plus activated oxygen).

However, the total Delfinaario design turned out to be too expensive and it was decided to cancel one of the show pool pumps and the tree fractionators it served, as well as the Photozone unit for treating the water returning to the pools. This change reduced the circulation to 600 m3/hr and the turnover to 2 hours 5 minutes(see fig. 3). Because of the elimination of the 3 fractionators the balance between fractionators and bio-filter was changed as well: 60% now went over the bio-filter and 40% over the fractionators.

This was the case at least on paper. What had been overlooked was that the nominal output of the pumps was for pumping the water to a height of 15 meters. In the Delfinaario system however, the maximum height for the bio-filter is 4.85 meter above pool level and for the fractionators only 1.75 meters. Except for the heat exchanger, no structures giving any resistance (such as pressure filters) were present in the flow lines. This resulted in a considerably higher output of the large pumps. The show pool pump that led its water through the heat exchanger turned out to be the only one giving 180 m3/hr. The other show pool pump and the holding pool pump both gave as much as 270 m3/hr. This meant among other things, that 3 of the fractionators got far more water than they should and were therefore not working optimally. With this in mind it was decided to rearrange the flow drastically. The show pool pump would still serve the bio-filter. The holding pool pump would now serve all 4 fractionators and by squeezing a valve its output would be reduced to 240 m3/hr. The water from the channel would be mixed with the water from the show pool and led over the bio-filter.

In the original design a separate section was made in the bio-filter for the medical pool, in case it had to be used as quarantine. This idea was later discarded, because of the risk of infection of the bio-filter with pathogenic bacteria which might become unmanageable. This separate section had been constructed anyway and now the water of the medical pool was led over that section in normal operation. If the need for quarantine should arise, its water would be led again over the fractionator as described above. The output of the holding pool pump would then have to be reduced to 180 m3/hr temporarily.

These changes resulted in a (forced) further reduction of the importance of the fractionators in the system : now 65% of the flow is led over the bio-filter and only 35% over the fractionators. The actual pump capacity is now 750 m3/hr and the turnover is 1 hour 40 minutes.

Click here to load the system design drawing, as it was originally built. Click here to load this drawing in a new window.

Back to Table of Contents

The biofilter

The bio-filter was originally designed to consist of 3 separate units, which will be treated individually in this chapter.

Trickling zone

In the trickling zone an array of sprinklers divides the water over the filter surface, which has an area of 27 m2. These sprinklers are located about 3.3 meters above the water level in the filter tank. The trickling zone is packed with a specially designed filling material, which allows for a lot of air space. There are several types of filling material possible. (Krüner and Rosenthal, 1983). The type used in the Delfinaario had not been used before, but it was believed to be a good and relatively cheap alternative. It consists of an array of large brushes, that look like giant pipe cleaners. These brushes are one meter long and have a diameter of 20 cm. The hairs of the brushes are made of a kind of plastic. The total surface area of one such a brush is over 3 m2. In the original design the brushes would be placed in three layers and the horizontal distance would be 20 cm. This means that there would be room for about 2200 brushes, which would give the trickling zone an internal surface of about 7000 m2.

Biofilter brushes

This design allows for a thorough aeration of the water because of the large air-water interface. An additional advantage of this kind of filling material is that the trickling filter cannot be blocked, as is possible with for instance stone-filled trickling filters (Mudrack and Kunst, 1986). The filter material resembles an enormous array of mini-cascades.

The filter material rapidly became overgrown with micro-organisms, which formed a so-called biofilm on it. As is normal in a trickling filter, the biofilm started sloughing, thus creating flocs.

It should be kept in mind that the filter was not built according to the above specifications. The building company placed the rows of brushes 30 cm apart instead of the specified 20 cm. So the material was packed less dense than designed. Also they installed only two layers of brushes, separated vertically farther than specified, instead of three layers close together. It took until June 1986, before the third layer was finally installed, partly submerged. Currently the trickling zone contains about 1500 brushes and has a total internal surface area of over 4700 m2.

Settling zone

In a trickling filter, a lot of flocs are formed. To remove them from the water the usual approach is to allow the water to flow slowly so that settling, or sedimentation, of the flocs is possible. In this system this is done by leading the water trough a large V-shaped tank. In this tank the water flows slowly, about 0.4-0.6 cm/sec. This is faster than in conventional settling tanks and therefore the settling was designed to be aided by two pumps that would take water from the bottom of the settling tank and lead it through two filters. In this way the sludge collecting on the bottom of the tank would be removed as well.

The filters would be shallow-bed sand filters that have an automatic backwash mechanism, which allows the filters to be backwashed without taking them out of operation. These modern, very compact filters are constructed by Aluko B.V.(1). These filters required a prefiltration with a maximum particle size of about 2 mm. It was assumed, that the pumps would be fitted with proper hairstrainers, so no additional pre-filtration was installed. However, the pumps were fitted with pre-filters which removed particles of 6 mm. or more, so larger particles than specified ended up in the filters and damaged the backwash mechanism. As a result of this the filters had to be taken out of operation in January 1986. It was decided not to have the filters repaired, but instead replace them with a multi-layered deep bed sand filters. This became operational in June 1987. This means that the system has been working without any form of mechanical filtration from January 1986 until the installation of the new filters in June 1987.

Click here to load the biofilter tank design drawing or click here to load this drawing in a new window.


Because the construction of the trickling filter was new, there was no absolute certainty to what extent it would work. There was no doubt that they would very efficiently aerate the water. As an extra safety measure a series of bioplates were installed. These plates consist of a frame of hollow slotted pvc-pipes, covered with an artificial fibre cloth. This cloth would give a large surface area for micro-organisms to grow on. The water was supposed to pass through the cloth, being cleaned in the process. Ideally the bioplates would be able to handle up to 5 m3/hr/m2. In this design they would only have to take about 1 m3/hr/m2. The operation of the cloth as a substrate for micro-organisms proved itself rapidly. Indeed, the organisms growed there so well that the pores of the cloth were blocked already after a few weeks of operation. It was very difficult to clean the plates effectively, because they had to be taken out of the filter tank and had to be taken apart for this. And even then they could not be cleaned properly. The result was that in very short time the bioplates became totally ineffective. At first this blocking meant that the flow over the bio-filter as a whole had to be reduced drastically, because all the water had to pass through the plates. There was no possibility to let the water flow past the plates and out of the tank as overflows were omitted during construction. Only after a few weeks of operation, such an overflow was finally installed, allowing more water to flow through the trickling filter, because the flow was no longer limited by the bioplates.

There were indications that in the plates anaerobic processes were going on, that caused the resuspension of organic matter, thus increasing the turbidity of the water. After making sure that the trickling filter was operating properly, it was decided on June 3, 1986, to discard the plates (and at the same time put the remaining trickling filter filling material in place).

Similar plates have been in use in the water treatment of the shark exhibit of the Noorderdierenpark in Emmen, the Netherlands. Because similar problems as found here occurred there as well, the plates were discarded in January 1986 (A. Buma, K. Klerk, J. van Duinen, pers.comm.)

Click here to load the system design drawing reflecting the system layout after removal of the bioplates or click here to load this drawing in a new window.

Back to Table of Contents

Foam fractionators

The foam fractionators in the Delfinaario were constructed by VRM(2). Each of them has a maximum capacity of 60 m3 per hour. The water is mixed with about 28 m3 of air per hour. The air is sucked into the water by the propelling motion of a specially designed impellor system, which at the same time stirs up the water and insures proper mixing. The air led into the water was originally enriched with Photozone, to enhance foam formation. However, it became clear, that because of a sufficient amount of surface active substances in the water, the foam fractionators were already effective without the help of Photozone. Because of the intense mixing with air the disinfecting effect of the ozone was minimal. Therefore in May 1987 the Photozone units were removed from their original places and moved to another place in the system (see the section on recent developments).

The intense mixing of water and air in the fractionators creates a dispersion of air bubbles in the water. These are allowed to rise to the surface (This is the factor limiting the capacity: if the water flow speed is too high, the air bubbles will not be allowed to reach the surface.). There, because of the organic matter present, foam will be formed. The foam will rise through a column that is narrower than the rest of the vessel. When the foam is rising, it will lose most of its water. The dry foam will spill over the edge of the column and is removed from there. The VRM Mk I fractionator needed a constant flow of fresh water to remove the foam. (The VRM Mk II fractionator has a special fresh water spraying and recycling device, which makes it possible to wash away the foam with a constant spray of fresh water, without using excessive amounts of water.) To improve the foam removal without using large amounts of fresh water, the foam collection space was enlarged and a spherical shape was attached to the lid, that would force the rising foam to the sides and into the collection space. In this setup rinsing the collection space once a day with fresh water is sufficient for proper operation of the foam fractionators.

Click on the image to load a larger version the fractionator design drawing or click here to load this drawing in a new window.

This figure shows the design of the foam fractionators. Left the origial design (VRM Mk I) and right as it is currently in operation at the Delfinaario.

  1. water inlet
  2. reaction chamber
  3. water outlet
  4. foam collection pipe
  5. foam collection area
  6. impellor motor
  7. impellor
  8. air intake pipe
  9. Photozone inlet
  10. air exhaust
  11. sprinkler system
  12. outlet to waste
  13. foam guidance device
The broken line indicates the water level in the fractionator

Foam collection in the top of a fractionator Foam fractionator VRM Mk I, unmodified

Foam fractionation is known to improve water clarity. This has been demonstrated in this system as well. If a fractionator has to be taken out of operation for maintenance for more than a day a reduction in water clarity can be observed. This is corrected as soon as the fractionator is back into operation. This also indicates that the capacity of the fractionator section as it is now is just about sufficient and that a change in the balance towards more fractionating capacity, as was originally designed, would be better. The capability of fractionators to remove micro-organisms from the water has been demonstrated by microscopic examination of the collected foam.

This contained high concentrations of diatoms and bacteria, while also occasionally ciliates and nematodes (usually dead) were encountered. Fractionators of a similar design but somewhat smaller have been used at the shark exhibit of the Noorderdierenpark Zoo in Emmen, the Netherlands. There they didn't work so well. The reason for this must be sought in the very low load of the system. About 2 kg of food fish were given there every 2 days (in a 400 m3 system) (J. van Duinen, K. Klerk pers.comm.). In the Delfinaario 30-40 kg of fish are given daily in a system that is only 3 times as large.

1) Aluko B.V., P.O.Box 125 , 3800 AC Amersfoort, the Netherlands.

2) Van Reekum Materials, P.O.Box 98, 7500 AB Apeldoorn, the Netherlands.

Back to Table of Contents

  Next page
Back to the Online Papers Page Back to the Publications Page