Some extensive background information to start

I'm attempting to reduce water consumption at 25+ production facilities, and I've identified that the largest consumption of water (60+%) comes from evaporation cooling for all the unit operations in the process. This is done at the facilities via an evaporative cooling tower. The towers are all the same base design, and relatively similar;

  • 2, 3 or 4 cells
  • counter flow induced draft (see picture)
  • design capacity 45-85 MW heat load, with most being 65-75.
  • the largest discrepancy is that each tower was designed for 10 degree Fahrenheit (5.5 C) Delta-T, but with a wet bulb reference temp that encompasses 350/365 days statistically per year for that respective latitude / longitude (facilities are spread out across the US Midwest, so that number varies). This is considerable, but shouldn't matter for what I'm about to ask (I include on the off chance it does) enter image description here

I understand the basic workings of a cooling tower (wet bulb temp, efficiency, cycles of concentration, make-up, blow down, drift etc) fairly well. With that, I understand that the amount of evaporation is proportional and fixed to the heat load. A such, ones first opportunity to reduce water is to recover/minimize blow down and drift. The facilities already have 0.1% drift on average, and run anywhere from 10-25 cycles of concentration in addition to recovering the blowdown for process water.

Despite this, the cooling towers still represent over 60% of water usage, so the next step is looking at the evaporation side. During the summer, the evaporative load is fixed and cannot be reduced. However, during the winter we already see reduction in water consumption. This stems from a combination of latent + sensible heat changes during the winter, as opposed to exclusively latent heat changes during the summer. As much as 25% of the heat load at certain facilities is dissipated via sensible heat change during the deep winter (January February). This is a commonly referred to phenomenon in the cooling tower industry and even has it's own correction factor in some simplified equations.

This has led me to believe that there is potential to control this effect and reduce water consumption even further. Adding to this evidence is the graph below. In this graph, one can see that for a substantial portion of the year, the basin temperature goes as low as 55 degrees F. The cooling water for the process does not need to be any colder than 70-75 degrees F.

Cooling tower basin temperature

Considering that the basin temperature is consistently so much colder than necessary, my gut reaction says "we have more cooling taking place than necessary". While we don't want to reduce total cooling MW's, one would want to minimize mass transfer driving force (to stop evaporation), and maximize temperature driving force (to increase cold air to hot water heat transfer). Doing this would shift more of the heat load from latent to sensible heat. I was thinking this could be achieved by slowing the circulation fans down with VFD's. Lower air flow rate for the same water flow rate, air gets saturated sooner to reduce evaporation, and the water circulating would be warmer to hopefully increase heat exchange between air/water. In effect, what I'm hoping will be the case is being able to operate the cooling towers as a "wet cooling" unit during the summer, but utilize it as a "partial dry cooling" unit when possible to save water, but without spending the capital to purchase an actual dry cooling unit.

Now my questions

  1. I cannot find ANYTHING online about doing this to reduce water consumption. Am I missing something blatantly obvious that shoots down the idea of trying to control the heat load towards sensible vs latent heat? (effectively trying to "dry cool" instead of "wet cool", but using a cooling tower designed for wet cooling)
  2. If the idea isn't completely bunk, how would I go about modeling this with mass & energy balances &/or a psychrometric chart to get a ballpark number for how much water could be saved? Given that I know: Historical wet bulb temp & temp, Hot in temp, cold out temp, fan amperage (er go fan speed), water re-circulation rate, blow down, make-up and cycles of concentration

2 Answers 2


I believe I have answered my own question after some time.

Can one control latent/sensible heat changes using climate? Yes, one can - but at a cost

while searching for more information on the topic I ran across a recent beautiful model for cooling towers that is easy to use. This model fairly well takes into account the behavior of water during the conditions input, while the only thing necessary is the operating conditions and a psycrometric chart.

The two operating parameters that I was changing were return temperature (return to the cooling tower, before being cooled) and the air mass flow rate (which I have reduced to fan speed %). By changing these and finding the appropriate humidity, saturation humidity and average inlet temp & humidity, I can model the cooling tower water consumption. pairing the cooling tower water consumption alongside the chang in sensible heat of the air from inlet to outlet (assuming identical to water inlet), I now have the total heat load on the cooling tower (sensible air + evaporative water).

With this, one can simply change the return temperature condition & input new humidity parameters, then run the model. The heat load is now not the same as the previous run, which is not comparable (heat load in the facilities is constant). This is corrected by adjusting the fan speed % (air mass flow rate) and running the model until I get the exact same heat load of sensible air + evaporative. I did this procedure for +/- 10 degrees F and I now have 3 sets of operating conditions modeled to compare.

enter image description here

The first thing that jumps out is that at higher return temps, we have less fan usage and more water usage. At lower return temps, we have more fan usage and less water usage. This is the fundamental "wall" that you have to deal with when using evaporative cooling. Now, these results make sense to me in hindsight for these reasons;

  1. If one looks at a pscyrometric chart, it can be seen that Saturation humidity goes up exponentially with temperature. This means the air can hold more water, and the mass transfer driving force is increased the hotter we let the air get, whith the inverse also being true
  2. This is analogous to dry cooling vs wet cooling - wet cooling uses less fan energy, but more water. Well, in this situation we're trading electric energy for water, just like wet vs dry cooling.

So in essence, if you allow the fans to run harder (and thus keep the cooling tower temperatures lower), you are maximizing the amount of dry cooling (sensible heat change), but using significantly more electrical energy. In contrast, if you slow the fans down, you use less electrical energy but increase your evaporation rate significantly (reduced sensible heat change).

There are some things I plan on looking into in the future when I run an actual experiment on some of these cooling towers. Most importantly, that changing the air flow rate linearly (fan speed %) changes HP usage of the fans as a cubic function, and the fact that the psycrometric chart saturation humidity increases exponentially with temperature; these things make me beleive there may be an optimization between electrical usage and water consumption here. Once I complete an experiment, I'll post the emperical results here to ensure the model conclusions match real world values.

Empirical trials - I had the opportunity to trial this idea at an industrial facility during the month of March. The trials were run at ~44 degrees north latitude. The setup consisted of 2 closely designed induced draft cooling towers. Both were 3 cells and designed to handle identical heat loads for the same weather conditions, although operations have changed and they were not handling even heat loads. Additionally, one tower is considerably newer than the other, and the fill media undoubtedly had an effect on the performance of the tower. To avoid this from detracting from the analysis, the older tower with less efficient fill was used as a "control" group. During two time periods that had average dry bulb temperatures of 33.7 F & 33.3 F (effectively controlling for temperature), the control tower was kept with a sump temperature set point of 65 F, while the variable tower was switched between 50 F & 65 F set points. The tower temperature controller was allowed to control fan speed to hit set point in both towers.

Cycles of concentration in each tower were controlled, as no changes to makeup water quality or conductivity set points were changed during the trials.

Heat load on the newer, variable tower averaged 38.6 MW & SD of 3.2. Control tower averaged 28.6 MW & SD of 3.6. It should be noted that MW estimation of the control tower is a bit crude, as I had to estimate flowrate based on the number of active pumps (as opposed to a live flow transmitter). Regardless, the method I used of relative changes in each tower to ensure climate effects aren't interpreted as an actual system response should stand.

The final caveat is that the colder sump temperatures of 50 F were difficult for operators to maintain over long periods of time due to water freezing issues on the louvers. This required frequent fan reversals, which bottled up heat in the tower and altered the effect sump temperature. This data was disregarded and only non-transient data at the achieved set point of 50F was kept for analysis.

On to the results: at the set point of 50 F, water consumption per MW of heat rejected was reduced by 16%, with the total fan kW consumption increasing by 168%. This falls right in line with the predictions of the model, and answers the question with a definitive Yes - one CAN control latent/sensible heat changes in a cooling tower by utilizing climate/weather effectsResults from trial

However, the true economic optimum will be variable dependent on water and electrical costs. Below is a graph showing the break even point of water vs electrical costs. Under certain conditions of extreme water costs (grey water reclamation, dirty water sources that require considerable pretreatment and chemical costs etc) and cheap electricity, it may make more sense to operate at the colder temperature. It should also be noted that this curve is for ~33 F dry bulb temperatures, and this will not be recognized all year long. Additionally, I anticipate that a different dry bulb temperature would likely change the slope of this line. As we increase the heat capacitance of the incoming air per lb (more available Delta T), we would require less air for a given heat load. This would further reduce electrical consumption for a given heat load. The converse would likely also hold true.

enter image description here


The fundamental problem you're running into is that evaporation was originally selected over sensible heating because it is much more energy dense.

When you evaporate water, you can reject 2257 kJ/kg. This requires a water flow rate equal to the heat rejected, Q, divided by the enthalpy of fusion, h.

When rejecting heat to water sensibly, however, you can only remove 4.18 kJ/kg-C. If we take a best case scenario where you can always reject heat to the winter water temperature, we can heat it by maybe 30°C before it starts evaporating really quickly. Here, the water flow rate must be equal to the heat rejected, Q, divided by the heat capacity times the delta T.

Comparing these two flow rates, you need 18 times more water to heat it sensibly. Then once all that water is hot, you need to cool it down again to reuse it.

There has been a lot of research into ways to do minimize the water consumption of water cooled condensers. In short, research has shown that you can't save a majority of the water through retrofits- you have to switch to dry cooling. That switch requires new equipment and hurts your system efficiency. If you want to look into the research, much of it was done in the concentrating solar power area since those plants are operating in the desert. This article isn't technical but may point you in some interesting directions.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.