Shrimp processing - raw material utilization.

[Guest David Lindsey]

I have got a problem on my hands that I have trouble finding a logical solution for. We are trying to find a way to get a continues read out of the raw material utilization in the shrimp processing plant. The plant is fully automated where the whole process from unloading the received containers of shrimps to packing the peeled and frozen shrimps in 0.4 kg plastic bags are don without the touch of a human hand. The basic structure of the plant is like this: 1. Shrimps are received in containers of 250 kg. 2. Preprocessed (cooked, salted etc) 3. Peeled 4. Washed 5. Blanched 6. Blowed 7. Inspected (optically and by laser) 8. Some may go back to 3 (peeling) after inspection 9. Graded 10. … 11. Frozen 12. Packed (0.400 kg plastic bags) We wish for the purpose of getting a better picture of the raw material utilization to place a weight at 1 (unloading of containers) and at 11 (before entering the freezer). This will give us the amount of unprocessed shrimps going into the processing plant and the amount of processed shrimps going out. Now we get to the problems The duration of the process is about 20 minutes, but a shrimp may go more then one round through the peeling machines so it can be 25-40 minutes for some and less for others. The plant never stops. We can’t weight the wastes/disposals How could we at any time get a more or less accurate prediction on the raw material utilization? The closest to an accurate answer that I can think of is the capture the data from the weight (flow-weight) at point 1 and store it in a data buffer and compare it to the data from the weight at point 11, with a 20 minutes time displacement. The amount of shrimps that will use more than 20 minutes on the process will the be an source of error. If any one has a better solution or some suggestions I would appreciate it very, very much.

[Ken Moore]

With this degree of automation, I must assume that you have a SCADA system or something similar. Couldn't you also calculate a through put rate? Pounds per minute entering the system, and pounds per minute leaving the system. Then historize and trend both values. You said you could not track the waste stream, but can you track the re-work(peeling)? Just a couple of thoughts. Ken

[Steve Bailey]

I don't see you getting precise data, only trends. When the plant is in a steady-state condition, the ratio of Lbs per out to Lbs per hour in should be reasonably constant. If it changes, the difference will be due to one or more of the following: 1. More or less waste from the shrimp. 2. An increase or decrease in the reprocessing rate. 3. An increase or decrease in the non-shrimp content (salt, water, etc) of the processed product. 4. An increase or decrease in the non-shrimp content (ice, water, etc) of the incoming 250 kg containers of shrimp. By tracking the incoming and outgoing weights over time, you may be able to determine characteristic patterns. For example, if you had one 250 kg container of shrimp that was particularly hard to peel, you would expect to see the yield ratio drop briefly and then ramp back up to normal. If that was followed by a 250 kg container with normal 'peel-ability', the yield ratio would increase briefly as the new batch worked its way through and then ramp down to the normal level. A sudden drop in yield followed by a sudden rise to normal would probably be the signature of a 250 kg container with higher than normal waste.

[Guest David Lindsey]

Hello Thank you for both the answers, they gave me more ideas on how to proceed. To Ken Moore: We have no SCADA yet, but the weighting issue has brought to our attention the possibilities that SCADA gives. But some of our current setup, gives us limited feedback from the machines, which means, we do not have the ability to get “condition data from the machines, like voltage, rpm etc.” So the SCADA has so fare been considered as overkill. But we have a fairly complex PLC solution. The newest of the machinery is able to give us more information which could be used in SCADA. What we wish to do with this weighting is to get a look at the material utilization and connect it to a set of parameters, some of who Steve Bailey mentioned, who has a great deal of impact on the material utilization. The ideal end result would be a computer program giving us a graph of the material utilization over time and at the some time showing the corresponding parameters. This could give us knowledge to determine things like: What is the optimal water temperature? What are the optimal settings on the peelers? What are the optimal settings of the lasers and inspectors? Only 1% better raw material utilization would mean quite much to us. To Steve Bailey: Yes I see that we will never get it perfect. You gave me the idea to place the weights in the process where all conditions will be the same on all weights: Shrimps will be unfrozen, and transported over flow-weights. One after 2. Preprocessed (cooked, salted etc) One after 7. Inspected (optically and by laser) and one before 11. Frozen Thanks for the help. Any further comment is appreciated.