### Need Help Designing an Experiment

I want to do an A-B comparison between two sauce pans. The experiment is to see which pan heats up a a batch of water faster.

Problem: the two sace pans are not the same size. Nor are they the same shape. But they are not that different.

Background: I want to test one of my Cuisinart multi-clad pots against a friend's All-Clad Copper Core. I've always been curious about the copper core and whether it is actualy more responsive than other types of S/S cookware. I was only able to borrow my friend's 2 qt copper core sauce pan for a limited time. The closest similar size in my MultiClad line is a 1.5 qt. (next size up that I have is a 2- 3/4 qt). The 2 qt All-Clad is a little taller and narrower than the 1.5 qt multi-clad.

I ran a few preliminary tests using the same volume of water (of the same temperature) in each pot on the same stove top burner (electric). The copper core brought the water to a rolling boil about a minute faster than the multi-clad. Just out of curiosity, I ran the same test in the larger multi-clad 2 3/4 qt size pot. This resulted in an even faster time than the copper core, which didn't make sense. There must be something about heating a larger pot with the same volume of water that heats the liquid faster. I thought maybe it's the height of the water relative to the bottom of the pan. So I went back to testing the 1.5 qt multi-clad against the 2 qt All-Clad copper core with the same level of water in each (2"). This time, the 1.5 qt was faster. But now I don't know what I've learned.

I need help designing an experiment that can simulate a direct A-B comparison between the two different size pots. Should the water be the same volume? Or the same level? Or should the water fill the same relative volume of each pot? Or is this whole experiement invalid without the same size pots?

What's the best way to eliminate variables and make the fairest comparison?

1. Click to Upload a photo (10 MB limit)
Delete
1. The problem is that it is difficult to compare two cookware of different sizes and shapes. Yes, you can get some results out of the experiment, but it will be difficult to give any meaning to it.

"I ran the same test in the larger multi-clad 2 3/4 qt size pot. This resulted in an even faster time than the copper core, which didn't make sense"

This is probably because the larger pot has (1) either a larger base, and therefore captures/absorbs more heat from the stove, or (2) a taller side, and therefore preserves more heat.

"Should the water be the same volume? Or the same level? Or should the water fill the same relative volume of each pot? Or is this whole experiement invalid without the same size pots?"

It all depends what is your question. All experiments start with a question in which you want to answer. If you want to answer the question: "which cookware has the greater heat transfer to the food, then you should use the same amount of water.

1. The fact that your two sauce pans are different capacity, I believe, introduces a variable that you have no way to normalize. Thus I believe you will never be able to set up conditions where you can make a fair comparison of which pan heats faster. I don't know the particulars of these two pieces of cookware, but typically a 1.5 qt pan will have both different diameter and height than a 2 qt pan, the same with the 2&3/4 pan. That different surface area on an electric hob will in itself change the way the pan heats. Water heats via the method of mixtures, that is, the water in contact with the hot part of the pan mixes with the water not in contact with the hot part of the pan, so the area of the bottom of the pan matters and the volume of water that needs to be heated matters. If you put 1 qt. of water in two pans of different sizes the one with the most surface area will boil first assuming they are both made of materials suitable for the task. In other words, if the two are similar enough the surface area will mask any difference in material performance.

The other big thing that you have to take into account is the thermal mass of the pan and the thickness of the bottom. A thicker pan may take longer to boil water because of the time it takes to heat the bottom, even if it has better heat transfer. Since boiling water doesn't take into account a very important factor in cooking, even heat distribution, a pan with low thermal mass will likely boil water quicker but not have even heat across the bottom that is necessary not to burn sauces. My guess is if you went out and bought the cheapest, thinest, 2 qt sauce pan you could find that was aluminum, no SS, it would boil water faster than either of the pans you are evaluating. That doesn't make it a better sauce pan.

The only way I know to make the fairest comparison is to use two pans that are as identical in size and shape as you can possibly find. And that will just be for those two pans and will not be applicable to other brands and shapes with different construction or thickness.

1. To follow up Mike's point. Technically speaking, you can set up to test which pan heats up the content faster. The real challenge is to get some bigger meaning out of it. In other words, pan A heats up water faster than pan B does, but it does not imply pan A has a more efficient design.

Mike is correct in term of the thin light cookware statement. Thermal capacity is definitely one reason. There is another reason as well. The heat transfer itself is greater with a thinner bottom (thus shorter distance). Thereis a balance to be made between heat response and temperature evenness, and there isn't a magic number to this. For a given design, you gain one, by losing the other one.

One more thing to consider for stainless steel cladded cookware. Stainless steel significantly affects the overall heat response by providing a high thermal resistivity. So, in many ways, the pan with the thinnest stainless steel walls will response to external heat the fastest. Yet, the cookware with the thinnest stainless steel is also the less robust and less durable.

At the end, every one will have slightly different critera. What is good for you may not be good for another. As long as your pan/pot is working fine for your style, then it is a good pan.

1. Hi, Seitan:

Having done a number of these experiments myself, I can tell you that it's all about doing your best to equalize all other variables. Unfortunately for you (and us, deprived of your results), pan size/surface area is one of the larger variables. If you don't have equally sized pans to compare, it is difficult.

Do you only have these 3 pans to compare? What are your results when you time boils with all the pans, say, 3/4 full? Starting from a different conceptual starting point, assuming in advance that the A-C and multiclad will perform identically, and if you have equalized other variables like hob size, you should see some linear relationship in the results. It follows that if you *don't* see something resembling linearity, you *may* be seeing a real difference that *may* be attributable to the construction difference. The problem with this approach is that it relies on inductive logic and in your case a tiny sample. If you had 4 pans of each line and the sizes overlapped (e.g., A-C in 3,5,7 & 9", and multi in 4,6,8 & 10"), you would have a larger sample, more data, and perhaps more reason for confidence in the conclusion you arrive at.

Still, if I read your OP correctly, I think I can make a couple of generalizations. It sounds as if your friend's A-C 2Q pan's footprint is smaller than your multiclad 1.5Q's. Correct? The fact that a smaller-footprint pan boiled a given volume of water faster (and by a minute) than did a wider-footprint pan I find to be significant. If your hob output was constant over the entire surface area, that result means that the A-C beat the multi by a minute and with *less* energy dumped into the A-C pan and its water. (This is no giant energy savings because the unoccupied area of the coil was still dumping energy into the room). I think it is a fair conclusion to extrapolate that an A-C with a surface area identical to your multi 1.5Q would boil the same amount of water *more* than a minute faster.

You might try this: Calculate the pans' bottom surface area in contact with the coil. Then adjust your volume of water so that you have the same volume to surface area ratio. This would be a fairer test. But be prepared for the A-C's margin of victory to be even greater.

Hope This Helps,
Aloha,
Kaleo

3 Replies
1. re: kaleokahu

Hi Kaleo,

I think the experiment with 2" of water in each pan was probably fairly close to what you suggest in the last paragraph. But in the experiment, the multiply pan was faster. Honestly, I think there are too many variables in the experiment to get accurate results.

Seitan, to get any level of accuracy, for starters you need to repeat the experiment a minimum of 5 times and preferably 10 times to get values with some level of reliability and they need to be timed as precisely as possible. Then you need to run the statistics to get the mean and standard deviation of your test. This data can then be ploted with error bars to determine if there are stastical differences in the numbers. You also need an accurate thermometer and bring the water to some temperature that is fixed, not a rolling boil which is somewhat subjective. You will need a procedure for testing the electric coils for start temperature and final temperature to insure the difference isn't in the coil and not the pan. Also important is where the pan is placed on the coil, so that each pan recieves the same contact area (within the limits of pan size) and same heat gradient across the coil every time, otherwise the standard deviation will be greater and void the experiment. This only leaves two variables, the one you want to test and the one that you are forced into with the availability of pans. That's a tough one and I don't know exactly how you are going to eliminate or compensate for that variable. Perhaps Kaleo's suggestion for a volume adjusted by the surface area of the bottom of the pan is as close as you are going to get, but even that may vary considerably with differing radii for the two pans where the bottom and sides meet. I know that looks like a lot of work, but if you want results that will stand up to review, then that's what it's going to take. We do a burn test and to insure reliability we use gas from a cylinder that has a specific burn temperature and we measure the flame height from the end of the burner, this is just the level of accuracy that is needed to conduct an experiment.

I'm still going to guess that the thinnest pan wins the boiling water test. It's the shortest distance from the heat source to the water and the water doesn't care about hot spots.

1. re: mikie

Very helpful suggestions. Thanks.

I see I'm going to have to cut down the margin of error even more. I'll use a thermometer instead of eyeballing a rolling boil. I'll use a larger sample size as well. Incidently, both pans cover the small electric hob. The A-C is about half an inch narrower than the M-C, but about an inch or so taller.

I'm also guessing that the smaller the pot, the less differences can be picked out from the margin of error. Bigger pots might yield more noticeable results from the tests.

I'll report back with updated results.

1. re: Seitan

You might also want to note how much each pan weighs. Put them on a food scale.

Use a measuring cup to measure the water volume not the height in the pan. Keep the volume of water a constant. But use more water, like 1 quart (4 cups) so you'll see a bigger difference.

Use the small electric hob and keep pan centered so the surface contact area is a constant. Use the SAME electric hob.

I think this experiment is very interesting. I have a hypothesis which may guide your experiments: the heavier pot with more heat capacity will boil in the shortest amount of time.

2. "What's the best way to eliminate variables and make the fairest comparison?"

You have variables all over the place. I don't see how you can eliminate them or even control them unless you are fabricating the pots yourself. The best you can do is to try to understand them. A big variable is the heat capacity of the pots themselves. That depends on the materials and the mass of each. The pot will soak up energy as it heats up. The more energy it can contain, the longer it will take to reach a given temperature.

1. re: GH1618

GH,

I was thinking that as well. On the other hand, there is a way to relatively minimize this by increase the volume of water. This will relatively reduce the heat capacity contribution to the total heat capacity. In other word, a 8 cups of water will tell a better story than a tablespoon of water. That being said, Seitan will need to form a null hypothesis in the first place; colloquial as a testing question. One design an experiment around a null hypothesis, not the other way around.

2. Another factor is geometry, as you acknowlege. However, if your "A" and "B" of you A-B comparison are the two particular pots, it doesn't matter. They are what they are. If, however, your "A" and "B" are two types of construction, then you need two pots of identical geometry and similar except for the thing you are interested in (presumably the copper core). The closest you can get to controlling that is to use two All-Clad pots, one the tri-ply SS/Al/SS and the other Copper Core, assuming that these come in exactly the same size and shape.

1. I've had a set back. The candy thermometer I'm using keeps getting fogged up and I can't get an accurate reading. I'll be looking for another thermometer. Feel free to post suggestions on type, brand.
In the meantime, I'll respond to some of the comments: Appreiciate everyone's input so far.

-My null hypothesis is that there will be no significant difference between the two pans
(both pans are clad with layers of similar metals [stainless, aluminum, etc, yet one pan has copper. I am testing to see whether the copper layer makes any significant difference).

-Thickness of pan is not necessarily a limiting factor on thermal performance. I ran a similar test between my M-Cs and a Demeyere Atlantis pot, which has a thick heavy disc bottom (see my post here on CH somewhere). The Atantis disc bottom pot blew away my multiclads.

-I won't be weighing the two pans, as this would be irrelevant, since the two pans are already different sizes and have different size, shape and weight of handles.

-Am using 4 cups of room temp water. Getting any more into the 1.5 qt m-c would leave little room for possible boil over

-Am using the same electric coil for both pans. Coil measures 5 & 7/8" diameter.

-Outside bottom measurements: 5.75" (all-clad) 5.5" (multiclad) Inside bottom measurements are difficult to obtain due to curvature of inside walls

-Once I resume tesing, I'll create a sample size of 10 (for each).

Hope to resume this weekend.

5 Replies
1. re: Seitan

"-Thickness of pan is not necessarily a limiting factor on thermal performance. I ran a similar test between my M-Cs and a Demeyere Atlantis pot, which has a thick heavy disc bottom (see my post here on CH somewhere). The Atantis disc bottom pot blew away my multiclads."

That's interesting, I believe it was sauce pans that Cooks Illistrated tested and commented on how long it took the disk bottom pans to heat up. The Atlantis is copper disk, so maybe that's not what Cooks Illistrated tested, although I thought there was a Demeyre in the test list.

I would suggest a thrmocouple with a remote thermometer, or you might get by with a quick read digital like a thermopen.

1. re: mikie

Yes, the Atlantis pots not only have copper as a layer, but silver as well

Thanks for the thermometer suggestion.

2. re: Seitan

"My null hypothesis is that there will be no significant difference between the two pans "

An ambitious question at hand. I would definitely narrow it down. Unless you have a defined question, it is difficult to set up a defined experiment. For example, it is easier to conduct an experiment with the null hypothesis of "high calories diet and low calories diet have no different impacts on cholesterol level in blood" than the null hypothesis of "high calories diet and low calories diet have no different impacts (period)" Is there a more specific question you like to answer?

If you are not sacred, then boiling off the water would further make the heat capacity of the pan less important, but it is an experiment requires greater attention.

Good luck.

1. re: Chemicalkinetics

You're right. I was being lazy and sloppy with my hypothesis.

Thus: There will be no significant (more than 60 seconds) differrence in the time taken to boil 4 cups of water at a temp of 90 degrees C (my city is at an elevation of 3400 feet) in either a 2 qt size clad pot made of stainless, aluminum and copper, or a 1.5 qt size clad pot made of stainless, and aluminum when heated on a 6" electric coil stove top.

1. re: Seitan

I see. In this case, this should be very easy to test, and I think you have already done a similar experiment. Just beware not to extrapolate too much from the end result. Good luck.

3. Just thinking about the materials in different clad pots and how they're arranged. I wonder if the all-clad master chef series with its outer and mid layer of aluminum would conduct heat faster? It seems that theoretically at least, when heat first hits the outer layer of a pan, stainless would transfer the heat slower than aluminum. The master chef series also only has three layers, compared with the copper chef 5 layers, allowing the heat to transfer through fewer layers of metal.

3 Replies
1. re: Seitan

Sure. Like we said earlier, a pure aluminum cookware would transfer heat faster than a cladded cookware. In a cladded cookware you are really buying a combination of performance, ease of maintenance, durability...etc.

1. re: Chemicalkinetics

Well, the mastechef series isn't 'pure' aluminum, it's still clad with interior stainless steel.

1. re: Seitan

Yeah, but my guess is that the MC2 has a faster heat response. This is assuming the interior (cooking surface) stainless steel layer is not much thicker than the All Clad stainless steel series, and it really shouldn't be. It comes down to the fact that the total stainless steel layer contribution of the MC2 is half of that of the Stainless steel series. This will improve the heat response.

My "pure aluminum" response is just as an example. A pure aluminum pan will response faster than a triply (stainless steel - aluminum - stainless steel) cookware. Thus, you can think of the MC2 as something in between and will heave the heat response -- also something in between.

2. By the way, the purpose of a copper core is not to boil water faster. It is to reduce the temperature differentials across the surface to compensate for the low thermal conductivity of stainless steel in pans which are clad on both sides.

2 Replies
1. re: GH1618

A very good point. To some extend, mikie has also mentioned this.

1. re: GH1618

I think the OP is seeing a pot shape and volume effect, not a material difference. I would expect the copper to make difference in the bubble pattern on the bottom of pot as it approaches boiling, especially when used on a small diameter burner. It should also matter when boiling a half inch or so of water. But with 2" or more, the thermal capacity of the water is significant compared to that of the pan.

2. Update:
I have a new thermometer and have been running trials between the two pots. I've gotten wildly variable times so far, so I've had to adjust some parameters. The starting temp of the water matters, as does the starting temp of the electric coil element. Using water fresh from the tap was too inconsistent, so I've had to switch to standing room temp water. Waiting half an hour or an hour for the burner to cool down isn't enough apparently, as there is residual heat remaining that is affecting the times and acting as a variable. I will have to wait much longer for the burner to come to a complete cool down in order to begin each time from a cold start.

So this is going to take longer than the weekend. Just wanted to give people waiting for results a heads up.

14 Replies
1. re: Seitan

Yep, those are prety much the variables I addressed above. And the variation in results is why you need to do so many tests and get averages. You could spend a fortune on test equipment to try to control the variables better, but for that kind of money you could jsut buy another pan or several. One problem with electric elements is they tend to cycle on and off, if they didn't, you could just set the pan on the hot element and not have to wait for cool down. That would be one advantage of a gas burner in this type of experiment. Another variable you might run into is that electric power is somewhat inconsistant and at different times of day or with other appliances running, there may not be as much power actually going to the heating element. Checking a hot element isn't going to be easy, as they get well above the capability of most temperature dection devices. Your kitchen thermometer isn't going to cut it for that kind of temperature.

Again, it's a lot of wrok and I'm not sure how much valuable information you will get although I'll admit, you have gotten my curiosity going, so don't stop now.

1. re: Seitan

"The starting temp of the water matters, as does the starting temp of the electric coil element"

Oh, hell yes. Make sure the pot has completely cool down before each of your trials.

"Waiting half an hour or an hour for the burner to cool down isn't enough apparently, as there is residual heat remaining that is affecting the times and acting as a variable. "

A bit tiny of residual heat should not matter too much because it should be small compared to the water heat capacity, but you definitely do not want the coil to be hot. You can always put another cold pot to cool down the coil, or a wet towel. I did these.

1. re: Chemicalkinetics

"Oh, hell yes."

Ok, I almost fell out of my chair! Fortunately I didn't have a mouth full of coffee. For those of us that make a living with science or have set up a number of experiments, this seems so obvious, but we are certianly not the majority.

If I were doing this I would use a pyrometer with a surface temperature probe to check the coil temperature when it's cold. The thing is these are not inexpensive. A Therma K from Thermaworks is about \$100 and a surface probe is almost another \$60, and this only takes you to 250 degrees C, not hot enough to measure a hot coil. And it would be nice to be able to measure that hot coil just to see how stable that part of the experiment is. But that would at least give you a good starting temperature and if one assums the coil is constant, a big assumption I know, then you would have eliminated that variable.

1. re: mikie

:D

Here is another thing. There is always some levels of standard deviation or uncertainty in an experiment. Maybe Seitan has already gotten the answer he is seeking for. Afterall, he wants to know if there is a significant difference between the All Clad and the MultiClad cookware in term of heat transfer.

If the boiling time has been 200 seconds +/- 15 second vs 210 seconds +/- 15 minutes, then maybe this is the answer -- there is no significant difference.

1. re: Chemicalkinetics

"Afterall, he want to know if there is a significant difference between the All Clad and the MultiClad cookware in term of heat transfer."

I'm not sure about that. The objective of the experiment has not been defined to my satisfaction, which makes the whole business seem pointless to me.

1. re: GH1618

How would the objective of the experiment be defined to your satisfaction?

1. re: Seitan

For example, note that Chem, above, has written that your purpose is to compare All-Clad to MultiClad cookware (as a general proposition). My understanding was that your purpose was to compare the two particular pans. If your experiment is well-defined, there should be no confusion on this point.

Then you need to state precisely what is being compared. Just saying that you are comparing rate of heating water doesn't do it. You need to specify how much water, starting temperature, final temperature, ambient temperature, heating method, measurement method, etc.

You need to identify all the possible sources of variation and explain how you have controlled them to minimize error. You need to show that the experiment is reproducible with consistent results.

You get the idea. I am hard to please when it comes to experimental rigor.

1. re: GH1618

Yes, my purpose is to compare the two particular pans as exemplars of two different brands of cookware employing slightly different materials in their construction. If I wanted to compare the overall cookware I would need to compare many pieces from each set. My purpose is to see whether copper, as a component of one type of clad pan makes any difference in how fast it brings liquid to a boil, compared to a similarly constructed pan that is also clad but doesn't use copper. But perhaps my main purpose is to see whether the A-C specifically is any different in this regard than my Cuisinart MC. If so, I may consider acquiring a few pieces of A-C copper core. If not, at least I can stop wondering about it. I've already done similar experiements with Demeyere Atlantis. The results were so stark that I stopped after 4 or 5 runs. The disk bottom design seems to be superior for this specific function at least. I'm just not that fussy about the Demeyere aesthetic form or the handles, so I didn't buy it.

Would love to rent a lab, have access to all kinds of instruments and larger, more equal size pans from each company, but I'm doing the best I can with what I've got.

I've outlined my parameters in an earlier post. I will reitierate these parameters in a full write up once my trials are over.

P.S. things are going well now I think. I've got the coil burner element under control and am getting an equal start to each test pan.

1. re: Seitan

You are trying to have it both ways. They do not serve your purpose of comparing the construction of the layers, because they have slightly different dimensions, which will affect your results. You would have to separate the effect of geometry from the effect of material. How will you do this?

1. re: Seitan

Seitan,

GH might have come in a bit tough, but his overall point is not invalid. This is something I tried to raise earlier. You can setup null hypothesis and you can setup an experiment, but one has to be careful about drawing bigger meanings than what the conclusion provides. I have seen so many people here made the this type of mistakes -- possibly due to lack of scientific or engineering understanding. I usually do not point out the the fact.

As I have mentioned earlier it is easy to set up an experiment to see which pot allows water to be boiled the fastest, but that does not really suggest anything about copper because really the stainless steel probably playing a bigger role.

http://chowhound.chow.com/topics/8323...

As for your comment about Demeyere disc bottom, this is very much valid. Few here have already pointed out that disc bottom can be better than full cladded cookware in this regard.

http://chowhound.chow.com/topics/7637...

2. re: Chemicalkinetics

"There is always some levels of standard deviation or uncertainty in an experiment."

That's why I suggested 10 repetitions with each pan. You can then get the average and the standard deviation and then plot error bars for the two data sets and determing if there is a statistical difference. Your example indicates there is no statistical difference, even though there may be a difference in the averages. Obviously, the tighter the standard deviation the more significant the 10 degree difference would become. That is naturally assuming 10 seconds is really significant in boiling a pot of water.

1. re: mikie

"That is naturally assuming 10 seconds is really significant in boiling a pot of water."

Now, there is another question, which is "does it practically matter?". Let's say we have a very well controlled experiment which give us very low deviations, and we deduced the boiling times to be 200.0 +/- 0.1 seconds and 210.0 +/- 0.1 seconds. That means the boiling time difference is only 5%. Thus, the time is STATISTICALLY significantly different, but not practically significantly different -- for me anyway.

1. re: mikie

An excellent point. One must know the error terms. A conclusion drawn from an experiment which is not reproducible is not valid. My guess is that the error terms in this experiment are rather large.

1. Which is the real question:
The experiment is to see which pan heats up a a batch of water faster.

I've always been curious about the copper core and whether it is actualy more responsive than other types of S/S cookware

Those are different questions. The first is more specific. 'responsive' in the second is vague.