gap
Apprentice
Posts: 390
|
Post by gap on Nov 21, 2010 16:18:47 GMT -5
Well put Sebastian
|
|
|
Post by jtoddm on Nov 21, 2010 21:51:21 GMT -5
Sebastian,
Thanks so much for all of your information on food safety. We are setting up a small bean-to-bar operation in San Francisco and have been going through our food safety plan and were wondering if you had any thoughts on the following:
On the one hand, the NCA/etc recommend 5-log kills, surrogate testing/validation, HACCP plans, firewalling your roaster so that the only way to get from the raw area to the clean area is through your validated control point, and reducing cross-contamination. This all sounds very reasonable and seems to be the best practice.
However, recently I was in france and spoke with a chocolate maker whose family has been making chocolate for the past 100 years. When I asked him about the 5-log kills, he said that if you achieve a 5-log kill, you will essentially be destroying a lot of the flavor. Like a good cheese, he believed that many of the bacteria are harmless and add to the taste (the "perfumes" as he liked to say). When pressed, he said that mostly they just take periodic samples, send them to a lab, and get the results back before any of the product reaches a customer.
I'm wondering if you think there is any merit to his point of view, and of doing random sampling? Specifically, I'm wondering if we can quarantine newly delivered beans, take a sample, and if they come back clean consider them safe? Also, since even a tiny amount of salmonella can be a problem, is sampling even effective?
Any thoughts would be appreciated. Thanks in advance!
|
|
|
Post by Sebastian on Nov 22, 2010 8:32:04 GMT -5
It's about mitigating risks. You'll never be able to eliminate them, but you can reduce the likelihood they'll occur. Just as wearing a seat belt doesn't ensure you'll not get hurt in a car crash, it does improve your odds. There certainly is an impact of roast on flavor - for many of the flavor grade beans, the preference it towards a low roast to preserve the volatiles. I've been to most chocolate factories around the world, and many of the old ones are using very outdated equipment that relies 100% on heat, in relatively large batches. In those cases - yes, i agree that if you're trying to hit the 5 log reduction via thermal processing and your mass is large, you're going to sacrifice flavor on some beans. I'd go one of two routes in those instances - either modify the roaster to allow for a steam injection up front, or reduce your mass per roasting batch to lower the thermal loading. Additionally, if you're bean roasting now, you could consider nib roasting, which requires a treatment to the bean that would lower the micro load and remove the shell prior to roasting - a step often referred to as micronizing. Using equipment that was created before salmonella was even identified probably isn't the most effective way of addressing salmonella It's all but impossible to get a micro read on beans themselves, and isn't going to be very useful. You'll want to do any testing on your liquor. I'm not a cheese maker by any stretch (although i do love it) i'm guessing there's a significant difference in the types of bacteria present. if anyone is advocating that salmonella are harmless bacteria, i think that should send up a red flag...he may believe that all he wants, just as i could believe i'm a professional football player - however at the end of the day, neither would be true regardless of how much we believe it to be. If any bacteria are present in quantities large enough that you can taste them, holy cow, he shouldn't be in business - and i doubt that that's the case. My suspicion is that he was referring to bacteria present during fermentation affecting the metabolites in such a way as to direct flavor. If i were you, I'd create a statistical sampling plan based on your batch size, frequency, and specific to each roaster/condition you have (i'm assuming there's one roaster and a small # of conditions). Again, since you can't sample 100%, it's about creating a plan that is a good balance to mitigate your risks. If you're tests are coming back indicating you've consistently got half a million TPC/g, you know you've got a problem. If you consistently attain much lower numbers, you're showing that you've got effective controls in place, and may wish to move to periodic sampling (quarterly, etc), as a monitoring program. Salmonella's a persistent bugger - it's very hard to eradicate once it's in your production equipment. If you're going with a stone grinder setup, frankly if found, your best bet is to simply get a new one. Edited to add - one more thing - i forgot to mention that the food laws here are much different than they are in other parts of the world, even in developed countries such as France. Need to be careful of viewing US legislation through the filter of practices in other parts of the world who have different - and usually less stringent - regulations.
|
|
|
Post by beanless on Nov 24, 2010 20:41:28 GMT -5
We are just looking at converting our garage into a small chocolate manufacturing plant and after reading the above I am wondering if I should have a separate room for sorting beans before roasting to prevent contamination? Does anyone else separate this process from the rest?
|
|
|
Post by Sebastian on Dec 1, 2010 11:53:08 GMT -5
|
|
|
Post by jtoddm on Dec 1, 2010 17:16:42 GMT -5
Sebastian,
As usual thanks for your incredibly insightful responses. A couple of follow up questions if you don't mind:
* Do you have any recommendations for information or classes on learning how to test for microbial load? We know of labs we can send samples to, but like most everything as part of this process, we'd much prefer to learn how to do this in-house.
* It sounds like you are concerned with the overall microbial load, but is it possible to simply test for the specific bacteria we are worried about? Is your thinking that if you can show that you have a low overall plate count at the end of the process then your other steps are effective? What would be considered a low/acceptable TPC/g?
* Similarly, how do you feel about the need for doing a surrogate test, i.e. loading up your beans with harmless bacteria before roasting and then determining how many were killed?
Thanks!
|
|
|
Post by Sebastian on Dec 2, 2010 17:45:14 GMT -5
Sebastian, As usual thanks for your incredibly insightful responses. A couple of follow up questions if you don't mind: * Do you have any recommendations for information or classes on learning how to test for microbial load? We know of labs we can send samples to, but like most everything as part of this process, we'd much prefer to learn how to do this in-house. --I would absolutely, unequivocally not advocate pathogen testing yourself. The last thing you want to encourage is growing diseases in a food environment. There are Adenosine Mono-Phosphate based detectors that operate on the basis - more or less - the more AMP is present is an indicator of how many microbes are present - add something to make AMP glow - use a handheld detector to read how bright the light is - tells you how many organisms in general you have. Useful for general indicators of sanitation, however not useful for determining if you have pathogens present. While it's an option, i would not likely recommend it unless you want to do routine environmental testing (which isn't a bad idea btw). Just a suggestion, but you might want to find a consultant - not just some schmuck on a web board - and identify your main pathogen risks (Salmonella, E. Coli), approach a local university to do the testing for you, or a local testing lab (Diebel, Covance, Etc). Your acceptance criteria should be absence of detection (ie no level of salmonella or e coli is acceptable). Make sure your testing is linked to your specific production protocols (record roaster type, conditions, temp, time, etc) so you can validate results to process.* It sounds like you are concerned with the overall microbial load, but is it possible to simply test for the specific bacteria we are worried about? Is your thinking that if you can show that you have a low overall plate count at the end of the process then your other steps are effective? What would be considered a low/acceptable TPC/g? -- See above. One thought would be to focus on main pathogens, based on a hazard analysis. TPC is useful for telling you if something in your process has changed or failed, but is not useful for pathogen determinants.
* Similarly, how do you feel about the need for doing a surrogate test, i.e. loading up your beans with harmless bacteria before roasting and then determining how many were killed? -- That's a very good suggestion. It's called a validation process, and you'd want to use analagous organisms that are similiar to the pathogens you're really interested in. In fact, you can use the same species, just a different serotype. Again, i'd stress not doing that yourself, but getting a micro-specialist to do it.Thanks!
|
|
|
Post by threechocolatiers on Dec 3, 2010 16:57:30 GMT -5
THANK YOU everyone, we've done several roasts now and each has been a great success. Much appreciate all of the input. Now we'd like to know about cleaning the Behmor 1600. The instruction manual suggests wiping the sides with Simple Green and then doing a dry roast, this should be done after 5 roasting cycles, does this apply more to roasting coffee beans or would it be the same for cocoa beans? We weren't sure if coffee beans give off more oil than cocoa beans and we don't want to ruin our roaster!!!!
|
|