The step of pre-flushing an espresso machine is a common procedure that for most, carried over from heat-exchanger (HX) based espresso machines. With the popularity of double boiler PID based systems, conflicting information exists as to whether pre-flushing to some level is necessary, and if so, how much. This post explores what affect pre-flushing has on a modern system: The ACS Vesuvius.
Some Background on Pre-Flushing Practices
Heat-exchanger based systems relied on the heat produced from the ultra-hot steam boiler to provide water at a temperature (approximately 94c/200F) that could be used for pulling an espresso shot through the group head. This is achieved via plumbing the brew lines through the boiler, heating the isolated brew water through thermal conduction. When idle, the design of the E61 allows for water in the group head to recirculate on its own through a convection-based thermosyphon action.
This approach, while simple and relatively effective, makes it more difficult to achieve the desired temperature for brew water in an HX machine. When the brew lever is not engaged, the water slowly recirculating in the group head will heat to nearly a boiling temperature. This would not only be far too hot for good extraction, but produce odd pressure in the group head because of its transition to vapor. The second issue is that water moving too quickly through the HX would not be capable of a high enough temperature for a sustained period of time. Too low of a group head temperature will decrease extraction yields, and increase acidity.
To work around this design limitation, a pre-flush expels the water too hot for a shot for a more suitable temperature. This is affectionally known as temperature surfing. With a group head temperature gauge on an HX machine, you can get extremely good at this practice to yield the temperature desired, and would often require a 13-16 second pre-flush to achieve the desired temperature, followed by a race to begin the pull shortly thereafter. I have a second machine (Quickmill Anita) that is an HX machine, and I could dial in the temperature really closely with that setup.
But the question remains. What about a double-boiler, PID controlled espresso machine? Surely a little pre-flushing is a good thing, right? Let’s take a look.
What to Test?
With the Vesuvius being a double-boiler, PID controlled espresso machine, I first needed to determine the criteria for testing. A large number of variables means that a thorough test becomes exponentially more complex, so I decided to perform my tests very similar to the way I use my machine.
The test itself was quite simple. Given a thoroughly warmed up machine (precisely 1 hour, with the bottomless portafilter in the group head), I was going to observe the group head temperature (via group head thermometer) throughout the pulled shot, and compare readings when pre-flushing (8 seconds), versus no pre-flushing. I’d repeat the process four times for each approach.
Test Settings for the Vesuvius
All settings on the Vesuvius were set to the default, with a few exceptions. The pump acceleration speed (set to 300) softens the transition from one pressure to another (thanks Dave Corbey) and the brew boiler set to 93c (199.4F) drops it down 1 degree from the factory.
As with other machines, the Vesuvius has a number of relevant settings that could be changed, such as an offset value to accommodate for the drop in temperature of water from the boiler to the portafilter basket. No other settings were adjusted other than what is noted above.
For this test, I used the following custom pressure profile. This has been a favorite of mine, as it introduces a soft pre-infusion for 14 seconds, runs at 9 bar, followed by a tail-off near for the last 10 seconds or so.
2 bar for 5 seconds
3 bar for 5 seconds
5 bar for 4 seconds
9 bar for 10 seconds
6 bar for 6 seconds
4 bar for the remainder
I use an 18g basket in my bottomless portafilter and aim for 36g of extraction. Due to the pressure profiling, my shots usually take much longer to extract than using a machine without pressure profiling: 34-44 seconds depending on conditions, with the first drip hitting at the 10 to 11-second mark.
The testing used the same batch of beans that I home-roasted. Extractions were done from day 3 to day 5 of their age.
For this test, I installed a group head digital thermometer right into the E61 group head. The thermometer does not give the temperature as the water hits the puck, with one manufacturer claiming a 3-5F degree offset, but it is certainly a lot closer than the PID used to control the boiler temperature. To get a more accurate group head temperature reading, one could use a Scace Thermofilter, but I didn’t want to purchase something that expensive just for this test.
I’ve had a lot of experience with group head thermometers, as I’ve always used them on my previous HX machine. Unfortunately, I grew tired of replacing the one brand that kept breaking on me, and only purchased this one for this experiment, and because it was a different brand from previous purchases.
Without any testing, it isn’t difficult to see that the method for controlling temperature on any PID based system is far from perfect. While a PID based system offers the control that is significantly better than a non-PID HX machine, it still suffers from a few key challenges.
- Measuring from the wrong location. For the PID to be able to control the temperature at the group head, it would require a much more complex solution to reduce temperature change as water passes through masses at different temperatures. Managing thermal mass can be tricky at best.
- Variable flow rates. Exaggerated even more on a machine capable of pressure profiling.
- No resistance in a basket is a poor way to observe the functionality of a PID. This is a big one. Watching the brew boiler temperature and the group head thermometer during a pre-flush shows a wide variety of temperatures. The PID has a more difficult time controlling the temperature at that high flow rate. A 15-second pre-flush will push out almost 4 ounces with pressure profiling because the resistance is minimal.
Depending on your espresso machine, your mileage may vary. The material used in the plumbing, how the plumbing is routed, the logic of the PID, and no pressure profiling could all contribute to very different results on other machines. Also, note that the small pre-infusion chamber that is a part of an E61 assembly is disabled on a Vesuvius. This allows the Vesuvius to have full control of the pressure profiling process.
The test results show an average temperature throughout the duration (36 seconds) of the pulled shots when comparing the two approaches of pre-flushing, versus no pre-flushing. Averages can sometimes suppress variation in the multiple test runs, so I also looked at the deviation on each run to ensure the averages were not masking anything.
There was less deviation in temperature when comparing pre-flushed pulls (2.52 degrees) to pulls with no pre-flushing (2.47 degrees). Although, the actual temperature results were very similar. The illustration also shows that even though the brew boiler temperature was more stable after pre-flushing, the group head temperature was not.
When comparing results of no pre-flushing versus pre-flushing at the time that extraction first hit the shot glass, no pre-flushing resulted in an average of 0.20 degrees (F) cooler, and closer to the target temperature than compared to the pre-flushing tests.
While not included as a variable in the test, I have anecdotally noticed that the group head thermometer reading tends to be slightly lower when the extraction is taking longer. This makes sense, as the temperature of the group head will always be naturally less (often times 194 degrees (F) when idle. These test pulls all ended at around 36 seconds, which is a bit fast for my setup. When it is closer to 38-40 seconds (more common), the group head tends to read a bit lower.
Also not tested was temperature stability due to the length of time for warmup. Due to the complexity of temperature control in an espresso machine, warming up your machine sufficiently is the simplest way to maintain temperature stability.
So there you have it. Given a sufficiently warmed up machine, the PID in my machine provides a temperature to the group head that has less variation of temperature at the group head than a pre-flushed approach. While you may wish to do a short pre-flush to clean the screen of remaining grounds, to warm up your coffee cup, or to satisfy your OCD tendencies, the practice of pre-flushing to establish the proper temperature seems unnecessary.
Great article. And the the others. Are you planning on publishing more?