Travelling: Thermos bottles tested (2019-4-8, and a few later updates, then some more updates in 2022)

some thermos bottles and bike bottles

Most recent updates:



I made 2 videos on the measurements and there you can see the bottles too (I will add standard pictures here though), on my youtube channel that contains mostly bicycle component reviews: my youtube channel:

Introduction, big thermos bottles (ca. 1 litre)

I've been using a cheap Aldi thermos bottle on my 2nd trip to Ukraine and I used that and later a 2nd version of the Aldi thermos bottle on all my trips abroad since that time, until I bought a Thermos King 1.2 litre. The Aldi thermos bottles work fine in a hotel or hostel, to make tea and drink it during the next hour. But I noticed that to go on a trip, and then make tea later, the sooner the better. E.g. From Budapest to Szentendre with the train, then a long walk, I felt the tea was not hot enough after an hour or 2. Another time for a night-time pic-nic in Zhytomyr (Ukraine) I decided to hurry up to make and drink the tea so that the water wouldn't cool down too much, so within ca. 1 hour. And that is about the point up to which this thermos bottle is still usable for that purpose...

In March 2018 I bought a Thermos 'King' 1.2 litre. I bought it after I had read a review on which showed (with measurements) how much better the thermos bottle 'King' from the brand Thermos was than various other thermos bottles. I used it on my next 2 trips, and it seemed to be better, but I had not done enough testing and I wanted some measurements so I did a test today. The advantage of better thermal insulation gets apparent not when drinking in a hostel or hotel, but when you travel with a thermos bottle full of boiling hot water, then put in a tea bag at your destination, or when you want to take a break from a long hike, then have some tea.

Introduction, small thermos bottles (ca. 0.5 litre)

I bought the 2 smaller thermos bottles for travelling to perhaps take instead of the 1.2 litre King, or along with it, but also for on the bicycle. They both fit in a standard bottle cage, not perfectly but good enough. Both have a diameter which is the standard 74mm of bicycle water bottles, but there is no indentation as with bicycle water bottles so they don't sit flush with most bottle cages at the end. In some types of cages they fit perfectly, such as the Point Retro cage with leather holder. In case of using one of them for making tea, I would not use the Active flask then, but instead the King 0.47 litre, but for on the bicycle I suppose the most likely use of them will be for cold drinks. This will come later this year... I am sure that both are a lot better than the Specialized Purist insulated, which works well for short trips (half an hour), but isn't a proper thermos bottle. Perhaps the Active flask, when using it with the tea/coffee cap or sports cap, will work well for drinking while riding. The King is definitely only suitable for drinking while stopped!

2019-4-16: Addition: I used to have an Elite Deboyo which is meant for cycling, with a stainless cap and a cap for cycling. I don't like any of Elite's drinking caps on their standard bottles nor on this thermos bottle, and as I didn't imagine using it further as it was hard to put into a standard bottle cage (contrary to a Point retro cage which I didn't use at the time, I think), I gave it away. However, today I measured it on a visit, so I added it to the table below.

2019-6-16: I tested a small 0.45 litre bottle, also from Aldi I think (bought ca. 2 years ago), cheap, but a lot better than the bigger bottles that I bought earlier. T-start=96.2 C, T-end (4h 4 min): 78.1 C, Tair 18 C, so thermal coefficient k=0.065 (the smaller, the better).

Summary of the test results, big bottles

There is a big difference of the cheap Aldi with the Thermos 'King': After 6 hours, in the 'King' 1.2 litre the water was still at ca. 84 degrees. I decided to end my test, and try to see how tea was going to taste and how hot it would be, made with water at that temperature. It was fine. After 2 cups I poured in the 3rd cup and measured the temperature (poured into the small cup that is also the top of the thermos bottle), which was 78 degrees. That is perfect for me as I like to drink tea at ca. 80 degrees, and the tea tasted fine as well with water of this temperature instead of made with boiling water.

The King 1.2 litre felt cool on the outside at all times, with a room temperature of 17-15 C, it was ca. 18-22 C. The Aldi bottle felt warm on the outside and in my tests I measured the difference, quite big!

The Stanley classic 1.0 litre is about the same as the Thermos King 1.2 litre in thermal properties (well, actually slightly worse), but ergonomically I like it better: There is quite a bit of rattle from the handle hitting the bottle with the Thermos King and little with the Stanley classic with handle folded in, none with the handle folded out, the inside of the handle is from a softer plastic which feels better, it pours better. It also has a longer warranty.

The Stanley quadvac 1.0 litre is the best vacuum bottle I tested so far in thermal properties.

To make a choice between bottles: I would go with a Thermos or Stanley, not some no-name brand as these bottles are really good in thermal performance. To then choose further there are a few considerations: The Thermos King 1.2 litre is the lightest (ca. 652g) and cheapest (ca. 35 euro) of the brand name bottles, with 5 year warranty. The Stanley classic is slightly worse in thermal performance, has a bigger opening (possibly useful for ice cubes? definitely useful for cleaning), has a nicer handle, water pours nicer into a cup, the cup feels less hot on the outside than the cup on the Thermos King, it is more expensive but has a lifetime warranty. The Stanley quadvac has a cloth like handle which means no noise in your backpack, it has the best thermal performance but is also the heaviest of the 3 at ca. 873 g.

Summary of the test results, small bottles

The small bottles performed quite well even if they were noticeably worse than the big thermos bottles, the Thermos King 0.47 litre is the best of the ones I tested.

The lid of active flask gets quite warm, it leaks heat there and that shows in the results when comparing it with the King 0.47 litre. The bottom of the Activeflask bottle is very well insulated, similar outside temperatures as the King 0.47 litre.

The Elite Deboyo is slightly worse than the Active flask, which I like more also because of the different caps included and because I don't like the bicycle drinking cap that comes with the Deboyo (sharp edges, which I could fix with some sanding, but also the dirt cap sucks and the shape is just not very nice to handle).

The use of small thermos bottles is for when you don't need much, which can be about short durations or to fit into the bottle cage of a bicycle. Many of them have a clearly inferior performance compared to the ca. 1.0 litre and bigger bottles from Thermos and Stanley. They are in most cases better suited to keep drinks cool (especially when using some ice cubes), all but the Thermos King 0.47 litre are not so good to keep water hot for say tea.

Specifications, prices

Small vacuum bottles:

Big vacuum bottles (ca. 1 litre and more):

Non vacuum insulated bottles for cycling:

Measurements of hot water over time in the bottles

The temperatures were measured using a Fluke 289 multimeter with a Fluke thermocouple (which has a low thermal mass, which works fairly well to measure even the outside of bottles with a small contact area between bottle and thermocouple.

I started measuring the outside of the bottles (to see where they leak heat) a few minutes after the test started. The low mass thermocouple is good here but also has limitations in contact area so some measurements seem a bit off and it took a while to get the thermocouple to reach the real temperature in most cases.

Thermos King 0.47 litre, Active flask 0.53 litre with standard cap, Thermos King 1.2 litre, Aldi 1 litre

Measurements on 2019-4-8:

    Boiled water poured into the bottles:
    Room temperature (Tair): 17 C at the start, dropping slowly to 15C (at night) after ca. 5h

    Thermos King 0.47 l | Activeflask 0.53 l  | Thermos King 1.2 l  | Aldi 1 l            |
    19:02:00 : 95.8 C   | 19:03:00 : 96.7 C   | 19:07:20 : 96.5 C   | 19:12:20 : 97.2 C   |
    outside:   18.6 C   | outside:   18.7 C   | outside:   18.2 C   | outside:   24.0 C   |
+1h:           92.3 C   |            87.9 C   |            94.4 C   |            83.0 C   |
    outside:   18.6 C   | outside:   21.0 C   | outside:   18.2 C   | outside:   27.5 C   |
                        | cap:       39.0 C   |                     | near cap:  33.5 C   |
+2h:           89.3 C   |            82.4 C   |            92.2 C   |            72.7 C   |
    outside:   19.0 C   | outside:   19.0 C   | outside:   18.2 C   | outside:   24.8 C   |
    near cap:  24.7 C   | cap:       36.8 C   | near cap:  22.8 C   | near cap:  28.2 C   |
+3h:           86.2 C   |            77.5 C   |            90.0 C   |            63.5 C   |
    outside:   18.8 C   | outside:   20.5 C   | outside:            | outside:            |
    near cap:  24.7 C   | cap:       34.4 C   | near cap:           | near cap:           |
+4h:       ca. 83.4 C   |        ca. 73.9 C   |            88.3 C   |            55.7 C   |
    (82.8 C at +13 min) | (73.1 at + 13 min)  |                     |                     |
+5h:       ca. 80.6 C   |        ca. 69.7 C   |            86.5 C   |        ca. 49.4 C   |
    (80.5 C at +3 min)  | (69.5 at + 2 min)   |                     | (49.9 at -4 min)    |
+6h:           77.8 C   |            66.4 C   |            84.6 C   |            44.8 C   |

And bonus measurement
+18h:          55.9 C   |

Specialist purist insulated bicycle bottle

Then we come to the Specialized Purist insulated bicycle bottle, which is not a proper thermos (vacuum) bottle, but good enough for short bicycle trips:

50 C water poured in the bottle:
Specialized Purist insulated (Tair = 12.5 C):
13:01:30 |  49.8 C
14:06:08 |  37.8 C (so at 1 h I estimate ca. 38.7 C with linear correction)
15:03:30 |  31.3 C (so at 2 h I estimate ca. 31.6 C with linear correction)

Elite Icy fly 650ml insulated bicycle bottle

In use it is easy to open the dirt cap, and to slide up the nozzle to drink. However, it is equally easy to slide it back down which happens by just touching it with yout mout, tongue, so this takes some getting used to.

Further experiences and measurement to come.

Elite deboyo thermal bottle with standard cap + bicycle drinking cap

2019-4-16: I measured the Elite Deboyo using the stainless steel cap (which is slightly larger in diameter than that of the Active flask, so the caps are not interchangeable), not the bicycle drinking cap:

Boiled water poured into the bottle:
Elite Deboyo 0.5 litre (Tair = 19 C):
start    |  95.0 C
+1h:     |  85.7 C
+2h:     |  78.5 C

2019-5-14: The Elite Deboyo's cap came apart. This happened after a not very long time of moderate use. The cap was stuck on one occasion, an underpressure was created in the bottle from cold water and the then contracting air. I got it open, then some days later the cap came apart. It had a tear in the stainless steel top... The cap is plastic with a stainless top and inside that it has 1 piece of foam that is badly cut and which does not completely fill the cap.

I tried to improve the thermal properties by using aluminium foil on the bottom, then a piece of roof insulation foam made to fit the cap exactly and glued it shut with 2 component epoxy. I measured the same way, each hour, to be able to compare to the earlier measurements (i.e. with heat loss from opening the cap each time). I was not sure if the aluminium foil would be useful, perhaps even counter productive due to the shape of the inside of the cap (it could conduct heat to the outside cap at the edges), I was thinking I should use it only at the very bottom or not at all, but I went ahead with all aluminium foil on which the foam was placed. The results:

Boiled water poured into the bottle:
Elite Deboyo 0.5 litre modified (Tair = 15.8 C):
start    |  96.5 C
+1h:     |  86.8 C
+2h:     |  78.5 C
+3h:     |  72.6 C

From this it seems that using the aluminum foil the way I did it was indeed a mistake, the results are slightly worse than the original version. I will leave it as it is, unless the cap gets loose at some point.

2019-5-15: I redid the test above of the modified Elite Deboyo, to see how reliable the measurements are and taking into account the slightly different ambient temperature they are almost the same.

Active flask with tea/coffee cap

2019-5-15: I tested the active flask with the tea/coffee drinking cap (not the sports cap) and as expected the results are slightly worse than with the stainless steel/plastic cap:

Boiled water poured into the bottle:
Active flask 0.53 litre with tea/coffee cap (Tair = 16.6 C):
start    |  96.6 C
+1h:     |  85.6 C
+2h:     |  77.3 C

Aldi another 1 litre bottle + a 0.45 litre bottle

Aldi 1 litre, earlier bottle: Tair = 18 C, Tstart = 96.4 C, Tend = 56.0 C, duration = 4.0944h, so k=0.177.

Aldi 0.45 litre (ca. 2017)  : Tair = 18 C, Tstart = 96.2 C, Tend = 78.1 C, duration = 4.066h, so k=0.065

Addition, 2022, Thermos King 1.2l bottle: loss of thermal properties

1 November 2022, after returning from my long trip through Poland and Ukraine I had been using the Thermos King 1.2l thermos bottle at home too and noticed tea got too cold after a fairly short time. Anything more than about an hour and it felt too cold (it will cool off to lower than the temperature in the bottle from pouring into a tea glass or into the cup that goes with the bottle, being colder than the tea you pour into it). Normally I drink all tea from the bottle within an hour when travelling, especially after cycling or walking a lot, but at home I drink a lot less quickly and so here it was quite noticeable. I also felt the outside of the bottle was warm, which it should not be... So I started a test:

  Room temperature (Tair): ca. 16 °C

  Water temperature in the 1.2l bottle which was fully filled:
     9:06: 96.6 °C
    10:06: 87.5 °C
    11:06: 79.3 °C

  Outside temperature near the cap after ca. 1.5 hours: 30 °C

So after 1 hour it is already at the point where originally it was after 4.5 hours... And the outside temperature is far above the ambient temperature whereas it was just 1°C above ambient after about an hour in my initial test.

This test confirms the bottle is no longer working as a thermos bottle. I don't know if this was caused by bumps the bottle will have received when it was inside my travel bag in buses etc. (though I always protected it with putting it next to clothes and towels or socks on the outside to protect it from knocks; there are no dents visible on the bottle so the knocks it will have received should not have caused an issue) but it is disappointing. On 2022-11-2 I sent an email to the company asking whether the 5 year warranty applies, which for the EU site seems limited to Germany. I bought it on but live in the Netherlands, so will the 5 year warranty apply? I got a reply on 2022-11-11 to send them the proof of purchase, so we will see what happens. [ 2022-12-9: Finally a reply and after reminding them that my address changed, they said they will send a replacement to my new address. ]

The replacement Thermos King 1.2l arrived. It is a matte (sand blasted or brushed) stainless steel this time whereas my old one was shiny stainless steel. I like this new look... Test started along with a retest of the Active flask, both with a 6 hour test. Temperature at the start was determined after letting the bottles sit with cap on for 1 minute to let that warm up the inside of the bottles:

  T room=15.0 °C
  Active flask:
  - start: T water = 96.6 °C,
  - after 6 hours (6h2minutes): T water= 66.0 °C
  k=0.0779 almost no difference from my original measurement...

  Thermos King 1.2 litre:
  - start: T water = 96.9 °C,
  - after 6 hours (6h2 minutes): T water= 86.1 °C
  k=0.0236 slightly better than the value of 0.027 that I calculated from 4x1 hour measurements in 2019...

So the active flask gives about the same results as using multiple openings. For the Thermos King 1.2l the difference is more clear, probably because it keeps the water hotter and thus gives more heat loss per opening.

Note about the active flask: It loses far more heat than really good vacuum bottles from Thermos and Stanly, and it has a large opening. This results in a low pressure inside the bottle after 6 hours (beause the hot air inside cooled down and thus has a lower pressure). I could hardly get the cap off! I had to get it to almost the top then pry it open diagonally, just rotating it off resulted in the cap just being sucked down again into the threads and not being able to remove it... Doing that multiple times should work too as there is going to be a little cold air that will get in, but I have not tested how many times you would need to do that.

Stanley classic 1.0 litre thermos bottle

This has a cool look I think, with the hammertone green. The cup gets less hot on the outside than the cup of the Thermos King 1.2 litre when pouring tea into it. It pours the tea a bit better, has a nicer handle (softer feel on the inside), and the handle rattles less. But it is heavier and more expensive.

With this one I will do 3 types of measurement:
1. Measurement with only temperature at the start and then at a 6 h interval
2. Measurement of the bottle lying horizontally, to see how much more heat loss you get this way with the hot water (instead of hot air) directly contacting the cap.
3. Hourly measurements to see how much each time you open the bottle for measurement influences the results, i.e. how much you lose in heat from hot air/water vapour leaving the bottle.

2022-12-4: Tests of the Stanley classic 1.0 litre thermos bottle:
1. Bottle standing up, 6 hour test:
Room temperature: ca. 13 °C, temperature at the start ca. 96.2 °C,
+6h: room temperature the same, water temperature ca. 84.3°C.
From this we can calculate k (the lower this is, the better).

Directly or from the formula to determine k: T(6)= 84.3= 13 + (96.2-13) e^(-k x 6) => -k x 6=ln((84.3-13)/(96.2-13)) = -0.1543 => k=0.0257. This is slightly better than the Thermos 1.2 litre bottle, however that was determined using hourly measurements. The influence of multiple times opening the bottle will be determined next. The outside of the bottle was ca. 18 degrees C.

Bonus measurement after 12 hours 40 minutes: I didn't leave the cup on the thermos bottle, which normally acts as insulator around the cap: room temperature ca. 13 °C, water is ca. 73 °C. (from this: k=0.0258 so this confirms the result of the first 6h measurement) This is borderline for me for tea. I did some testing again of at which temperature I like to drink tea. 75 °C is about optimal, 70 °C is the minimum acceptable as below that it starts to feel too cold. Note that you lose some heat energy from pouring the tea into a cup which needs to be heated up by the tea, so for me the water should be at least 75-80 degrees when pouring it into a cup.

2. Measurement with bottle lying horizontally, 6 hour test:
This has just about no effect, pretty good, I expected more heat loss.

Start: room temperature = ca. 12.0 °C, water = ca. 96.3°C
+6h: room temperature = ca. 12.3°C, water = ca. 84.1°C,

So barely any difference, from the formula: k=0.0261

Bonus measurement after 24 hours 10 minutes: I left it standing without the cup on the thermos bottle, which normally acts as insulator around the cap. I also noticed that I had not screwed on the stopper tight, it was slightly loose which will have caused a bit of thermal loss:
room temperature ca. 13 °C, water is ca. 58.4 °C. (from this: k=0.0251)
This is about expected, according to the calculations using k=0.0257 (for standing up, with cup on the bottle), it would have been at 58 °C after 24 hours. It looks as if leaving the cup off has almost no influence...
Note: The relative differences lie in the range of measurement/deviation error as the thermocouple measurements fluctuate a bit... (about +/- 0.3 °C, I then take the middle of the range as the measurement value). See more on measurement accuracy further on.

3. Measurement in 4 intervals of an hour:
This is to see how much opening the bottle each time influences the heat loss. Start: room temperature = 13.0°C, water temperature = 97.0 °C [ the bottle was still warm from the previous test, i.e. water of 60°C, so it was preheated, hence a higher initial temperature of the water than in the previous 2 tests.
Note: If the water temperature is higher it likely indicates that the bottle's internals didn't warm up yet, in that case the drop in temperature in the first hour will be a lot higher than the next trop. If so the first measurement should be discarded. So for the first measurement to be useful it needs to done a little while after pouring in, a minute or so, for more accurate results.. This is for example the case with the Active flask measurement, I should redo that though it doesn't matter much, it is not that great for keeping drinks warm anyway. I use it only to keep drinks cool. ]
+1 h: 94.5
+2 h: 92.0
+3 h: 89.5
+4 h: 87.7
Note also that the measurements from the thermocouple are not actually accurate up to 0.1°C, and they drift around a bit, which explains other seemingly inconsistent deviations such as the differences above per hour are 2.5, 2.5, 2.5, 1.8, which means possibly the measurements at the start are slightly too high. In any case, the results from the first and last measurement give: k=0.0293, a bit worse than the Thermos King 1.2 litre. I will later check what the results are of calculating k using linear regression. In any case the results show slightly worse results than opening once. The 4h temperature should be about 88.7 °C for a k of 0.026, and 1 degree is far more than how much the thermocouple fluctuates.
Notes about measurement accuracy: The fluke thermocouple has a stated accuracy of "+/- 2.2 °C or 2% whichever is greater ((0 to 260 °C)", but this is about absolute deviations. There are 2 deviations: we are using values that are all measured with the same thermocouple and thus deviations are about the same in all cases of measurements of a given temperature, with all measurements of say 90 degrees the absolute deviation will the same. There is some variation when measuring a certain temperature from various causes, but the thermocouple values don't just change all the time by 2.2 degrees up and down, the fluctuations are far smaller. Btw. comparing to another thermocouple that camse with another multimeter the temparatures are the same.

Stanley quad vac 1.0 litre thermos bottle

2022-12-13: I ordered the quad vac 1.0 litre on 5 December, it arrived 23 December. This should have noticeable better thermal insulation (keeps hot water 35 hours hot vs. 24 hours according to Stanley's own information), from amazon USA as I couldn't find it on, and on it was only available at ridiculous prices. It was on offer so I only paid about 45 euro including postage and EU(NL) VAT.

Height: 32.5 cm, diameter 9.5 cm, the opening on the inside below is smaller than at the top, that smaller opening has a diameter of ca. 38 mm, a bit smaller than the Stanley classic 1 litre which has an opening of ca. 44 mm. The cup is a thermos cup... It doesn't have a fixed handle but a handle from a plastic 'cloth' that wraps around the bottle which you can can remove. This means no rattling noise from the handle which is a pluspoint when walking with it in your backpack.

6 hour measurement, with no pre-heating of the bottle, I waited 1 minute to let the bottle's inside completely warm up before the initial temperature measurement:

  In progress (2022-12-23):
  Stanley quadvac classic 1.0 litre: 
  T room = from 14.4 °C to 15.2 °C,
  - start: T water = 96.0 °C,
  - after 6 hours: T water= 86.8 °C
  from this, k=0.020

It is better than the ca. 86.1 °C of the Thermos King with a lower outside temperature and slightly lower initial temperature which compared to the test of the first Thermos King 1.2 litre makes clear that this result is very good. The calculation of the thermal coefficient shows it: ca. 0.020, the best of any thermal bottle I tested so far.

Other considerations: It is a lot heavier than the Thermos King 1.2 litre at a weight of 873 g vs 652 g. The cup is part of this, it is the best of all thermal bottles too, not too much of a thermal cup which can cause your tongue to burn, and yet keeps the drink a bit warmer. It feels good i nthe hand with a ridge for your fingers to keep it from slipping, weight of the cup is 115 g vs. 78 g for the cup from the Stanley classic 1.0 litre. The cup of the Thermos King weighs xx g.

To do:
1. I will do a further test as I may not have screwed on the top tight enough as I thought the edge of the cup on the bottle was slightly warm.
2. Try the cup.

Bonus measurement: Another +14 hours later, so 20 hour measurement: the water dropped (from 86.8 °C that the water was at after the first 6 hours) to 70.6 °C. Room temperature is about 14.5 °C, the thermal coefficient k is thus approximately: k=0.018. Even better. I will redo the first measurement just as a confirmation.

Bonus measurement 2: Another 15.5 hours later, so 35.5 hours, about the same that Stanley says that this bottle will keep drinks hot (35 hours): T room is still about 15 degrees, T water is 58.0 °C.

How to measure thermos bottles

You don't really need to make many measurements as I did the first tests. From physics it's clear that you only need 2 measurements, at the start and a 2nd one after which the temperature dropped noticeably (for accurate estimation of the thermal coefficient), say 4 hours. That will give a more accurate value of k (no heat loss during each opening). Note also that the temperature dropped during the measurement from ca. 17 °C to 15 °C end of the measurements and during the night and morning it was ca. 12 C. To eliminate such variations it's better to measure during the middle of the day when the ambient temperature stays approximately constant. A duration of ca. 4 h is likely better than a longer period, to avoid changes in the surrounding air temperature. When doing it in the daytime, say. from 10 to 18 temperature should not change much, so a 8h test could then be useful esp. to more accurately measure the differences in performance of the bottles with good insulation. Using multiple measurements does have the advantage of being able to spot measurement errors as yuo can do an interpolation or regression to find a better value for the thermal coefficient k.

Thermal calculations:

The heat loss can be described by an exponential equation because the heat loss at any time is a factor of the temperature difference of the water and the surrounding air, i.e. the bigger the difference, the bigger the heat exchange (not taking into account phase changes of the water that might happen, i.e. melting ice cubes or the water freezing if the outside temperature is below zero): during a small amount of time, delta T = -k (T-Tair)

A quick derivation to show what goes on:

So : T(t)= Tair + (T0-Tair) e^(-k t).

Calculating the thermal coefficient k: From this formula we can calculate k from 2 temperature measurements with a given interval t: k= -ln( (T2-Tair)/(T0-Tair) ) / t, with t the time in hours.

k is the heat transfer coefficient, which can be calculated from the measurements. From the measurements I did, it is best is to take the measurements after 6 hours, then check the in between measurements. Or do an exponential regression! The variation in air temperature doesn't seem to make much difference to the results. The results of the Activeflask don't fit as precisely as the others, not sure why, perhaps due to the large opening to measure which will give more heat loss at the time of measurement. The biggest error in measurements comes from the thermocouple (value varies for a little while, then usually stabilises but sometimes it stays a bit unsteady).

The measurements of thermal coefficients using 1 hour measurements can have some deviations from opening times which can vary per bottle, but relatively show what goes on. Also they do have the advantage that you can do exponential regression on them (which can be doen the same as linear regression).

My preferred method since the initial tests is to pour boiling water into the bottle, put the lid/cap on, let the bottle warm up for 1 minute, open it and measure the initial temperature point, then do a measurement 6 hours later.

Thermal coefficients calculated from hourly measurements or a sinlge long duration measurement:

Small bottles:
- Thermos King 0.47 litre:                 0.043 from 4x1 hour measurements
- Aldi 0.45 litre (2017):                  0.065 from 4 hour measurement
- Active flask 0.53 litre:                 0.080 from 4x1 hour measurements
- Active flask 0.53 litre:                 0.078 from 6 hour measurement
- Active flask 0.53 litre, tea/coffee cap: 0.138 from 4x1 hour measurements
- Elite Deboyo 0.5 litre:                  0.120 from 4x1 hour measurements

Big bottles:
- Thermos King 1.2 litre:                  0.027 from 4x 1 hour measurements
- Stanley classic 1.0 litre:               0.029 from 4x 1 hour measurements
- Stanley classic 1.0 litre:               0.026 from 1x 6 hour measurement
- Thermos King 1.2 litre:                  0.024 from 1x 6 hour measurement
- Stanley quad vac 1.0 litre:              0.020 / 0.018 from 1x 6h measurement, 1x 14 h measurement
- Aldi 1 litre:                            0.180
- Aldi 1 litre (older):                    0.177 from 4h measurement
Not a proper thermos vacuum bottle, but somewhat insulated:
- Specialized Purist insulated bicycle bottle: 0.350

The Thermos King bottles are a lot better than the others... The Elite Deboyo is significantly worse than the Active flask which could be because of the even bigger opening on top and more heat loss there.

By calculating the temperatures at the measured times, you will find that the measurements are mostly accurate calculating with this equation (deviations fall into the accuracy, so possible deviation, specified for the thermocouple). So a long term measurement of 4 or 6 hours along with the initial temperature is really enough to calculate k and from that you can calculate the temperature at any time.

From this result you can calculate what would happen with 100 deg. C water in the Specialized bottle: after 1 hour at 17 deg. room temp. it will be 17 + (96-17) e(-0.35) = 72.7 degrees. Which is respectable but already too cold for tea.

The effect on cold water can also be calculated: Suppose we have a refrigerated drink, T0 = 5 degrees, and it's hot outside, with Tair = 35 degrees, how cold will the drink stay after a few hours?

E.g. with the Specialized bicycle bottle: 1h: 35+ (5-35) e^(-0.35) = 13.8 deg. C

For all bottles: temp. in deg. C, assuming no heating up of the water from the inside of the bottle...:

                        | 30 min| 1hour| 2 h  | 3 h  |   | 8 h
Thermos King 1.2 l      |   5.4 |  5.8 |  6.6 |  7.3 |   | 10.8
Aldi 1 l                |   7.6 |  9.9 | 14.1 | 17.5 |   | 27.9
Thermos King 0.47 l     |   5.6 |  6.3 |  7.5 |  8.6 |   | 13.7
Aldi 0.45 l             |   6.0 |  6.9 |  8.7 | 10.3 |   | 17.2
Active flask 0.53 l     |   6.2 |  7.3 |  9.4 | 11.4 |   | 19.2
Elite Deboyo 0.5 l      |   6.7 |  8.4 | 11.4 | 14.1 |   | 24.4
Specialized Purist ins. |   9.8 | 13.8 | 20.1 | 24.5 |   | 33.2
(ca. 0.63 l)

So after 3 hours only the Thermos and Activeflask bottles keep the drink actually cool, 17.5 C and 24.5 C isn't really cold enough to feel refreshing even when it's 35 C. The Elite Deboyo that I later added keeps it at ca. 14.1 degrees after 3 hours which is also not that good. Note that in actual cooling down of the body (when bicycling or doing other sports) there is not much difference between warm and cold drinks, sweating does a lot more! I calculated this on my bicycle site in my review of various bicycle bottles (at the end of the review). The main purpose of cold drinks is to get a feeling of cooling down even if in reality it doesn't do much...

Pictures to come. Video is on my youtube channel.

Thermal calculations from manufacturers' specifications

What does it mean if a bottle is said to keep drink hot for x hours? The limit seems to be set to 65 degrees by various manufacturers, which is too low for me for tea, but from that we can calculate at what duration the water temperature will drop below 80 degrees.

Assuming an ambient temperature of 18 C, and that the boiling water poured into the bottle becomes ca. 97 degrees: 65 = 18 [ Tair ] + (97-18) e^(-k x), so (65-18)/(97-18) = e^(-k x), so ln((65-18)/(97-18)) = -kx, so k= -ln((65-18)/(97-18))/x. x=24 (hours), so k=0.0216 which is lower than what I measured. Then to this question: at which time does T become 80 degrees? 80=18+(97-18) e^(-k y), y= -ln((80-18)/(97-18))/k, so y = [ ln((80-18)/(97-18)) ] / [ ln((65-18)/(97-18))/x ] (hours)

Suppose a bottle is said to keep drinks warm for 24 hours (such as the Thermos King 1.2l), so x=24, then the duration at which the temperature has dropped to 80 degrees C is: y = [ ln((80-18)/(97-18)) ] / [ ln((65-18)/(97-18))/24 ] = 11.2 (hours).

From my calculation of the coefficient k of the Thermos King, does this give the same result? No, my measurements give slightly worse results. The Thermos King 1.2l should have the water after 24 hours at 18+(97-18)e^(-k 24) = 59.3 degrees, and after 11.2 hours the temperature will be 76 degrees (C). This is close so their claim is possibly with a higher ambient temperature or there may be sample variation, and so then it may be at 65 degrees after 24 hours and 80 degrees after 11.2 hours.

The data from manufacturers are likely 'optimistic' but using that data as per the above gives a good indication of actual thermal properties, not meaningless statements such as 'keeps drinks hot for 24 hours'.

Now look at another bottle from Thermos, "THERMOS 4045.232.100 Ultralight Thermos Flask, Black, 1 L, Insulated Flask with Cover and Carry Strap", which is 1l with a large opening and said to keep drinks hot for 10 hours. The big opening and heat loss at the cap are likely the cause of that difference (Thermos King 1.2l: 24hours). So, how long would it take for the water to drop to 80 degrees? y = [ ln((80-18)/(97-18)) ] / [ ln((65-18)/(97-18))/10 ] = 4.7 hours. This is still fairly useful. The heat coefficient is then at best: k= -ln((65-18)/(97-18))/10 = 0.052.

Conclusions for the tested bottles:

Conclusion for the big bottles:

Conclusion for the smaller bottles:

General conclusion on cheap/expensive bottles, and type of bottles to use

For big thermos bottles that you want to use to make tea, get a good brand, esp. Thermos and Stanley are very good. They will also work better to keep cold drinks cold in summer, but for this purpose I like to have a bottle with bigger opening so the entire bottle can be cleaned easily (which is not really needed if you use it to keep tea in nor if you use it to drink cold water, but if you use hot chocolate milk, cold juice or cold sports drinks then cleaning becomes essential).

I'm still considering options. What would be better?

For bicycling:
- 2 x 0.5 or 0.7 litre thermos bottles on the bike in bottle cages,
- 1 x 0.5 or 0.7 litre thermos bottle in a bottle cage, and along with that use 1 or 2 bigger thermos bottles (1 or 1.2 l) in the rear bags when making long bike trips and then pour from the big into the small bottle when the small one is empty.

For hiking:
- 0.7 litre thermos in the side pocket of the backpack, or just put a 1.2 litre Thermos King there? That just about fits... But the small bottle + big bottle strategy as with bicycling may also be an option.

Other tests to come

The bottles were all tested with opening multiple times, it shouldn't affect the results much but I will test with a 4h and/or 8h interval. Also they were all, in my tests so far, tested standing up, but what about horizontal? That should give worse results as the hot (or cold) liquid touches the end cap which is the main point of where good thermos bottles leak heat. The normal use is upright, in a backpack with other stuff, in the side pocket of a back pack, and upright or 45 degrees in a bicycle bottle holder. Horizontal is rare, for short durations lying your backpack on the ground so testing this is not very important. If the bottle is nearly full, will the water touch the cap at 45 degrees? If so then it would be useful for that case, and that is more likely with bottles with big opening and thus wide lids, such as the Active flask and Elite Deboyo.

So I'm going to measure the King bottles and the Active flask, + perhaps a few others (see further) in the following way:
- 1. Pour boiling water into the bottles, measure temperature then, leave it standing up (typical for use in side of backpack or on bottle holder of a bicycle), and measure again after 4 hours.
- 2. Pour boiling water into the bottles, measure temperature then, leave it lying, and measure again after 4 hours. For good thermos bottles at least 6h is better as it gives a higher accuracy due to low drop of temperature. [ Done for the Stanley Classic 1.0 litre ]

Other bottles that I may try