SQL return limited rows based on agregating sum - sql

Question

I want to return a number of rows from one table whose sum is dependent on a value from a row in another table:
Scenario: Sales order for a qty of particular item. The item is found in a number of Bin locations. The storeman needs to be directed to the oldest material.
I can create a query that will list the Bin, the Qty in the bin and list them in age (oldest to youngest) - all good so far, but say the order is for 100 units and there are 50 or so units in each bin and there are 40 bins, then I don't want to list all the bins, just the oldest two - just enough to be able to fulfill the order.
How do I do that?
Just some more info as requested
DB = MS SQL 2016
Sample Data:
The following is the data for a particular item showing the Bin, the qty in that bin and ageing date:
Bin#, Qty, Date
1,40,2018-05-15
3,45,2018-05-15
8,45,2018-02-10
12,45,2017-11-11
13,45,2018-02-10
15,45,2017-09-02
18,20,2017-09-02
The sales order is for 100 of these items, We want to pick FIFO (First-In-First-Out), so the results I want to return are:
18,20,2017-09-02
15,45,2017-09-02
12,45,2017-11-11
These three bins contain a total of 110 units so that is enough to satisfy the Sales Order. Note that order is Date, then Qty
The actual query is currently:
select
[OrderHed].[OrderNum] as [OrderHed_OrderNum],
[OrderRel].[OrderLine] as [OrderRel_OrderLine],
[Part].[PartNum] as [Part_PartNum],
[Part].[PartDescription] as [Part_PartDescription],
[OrderRel].[OurReqQty] as [OrderRel_OurReqQty],
[PartBin].[BinNum] as [PartBin_BinNum],
[PartBin].[OnhandQty] as [PartBin_OnhandQty],
[PartLot].[FirstRefDate] as [PartLot_FirstRefDate]
from Erp.OrderHed as OrderHed
inner join Erp.OrderDtl as OrderDtl on
OrderHed.Company = OrderDtl.Company
and OrderHed.OrderNum = OrderDtl.OrderNum
inner join Erp.OrderRel as OrderRel on
OrderDtl.Company = OrderRel.Company
and OrderDtl.OrderNum = OrderRel.OrderNum
and OrderDtl.OrderLine = OrderRel.OrderLine
and ( OrderRel.OpenRelease = True )
left outer join Erp.PartBin as PartBin on
OrderRel.Company = PartBin.Company
and OrderRel.WarehouseCode = PartBin.WarehouseCode
and ( not PartBin.BinNum like 'Q' )
inner join Erp.Part as Part on
OrderDtl.Company = Part.Company
and OrderDtl.PartNum = Part.PartNum
right outer join Erp.Part as Part
and
PartBin.Company = Part.Company
and PartBin.PartNum = Part.PartNum
inner join Erp.PartLot as PartLot on
PartBin.Company = PartLot.Company
and PartBin.PartNum = PartLot.PartNum
and PartBin.LotNum = PartLot.LotNum
where (OrderHed.OrderNum = #SalesOrder)
order by OrderDtl.OrderLine, PartLot.FirstRefDate, PartBin.OnhandQty

Answer 1

You can select the bin where their date is less then or equal the minimum date for which the sum of the quantity of all bins with a date less than or equal is greater then or equal your target quantity (e.g. 50).
SELECT *
FROM bin b
WHERE b.date <= (SELECT min(bb.date)
FROM bin bb
WHERE (SELECT sum(bbb.qty)
FROM bin bbb
WHERE bbb.date <= bb.date) >= 50)
ORDER BY b.date,
b.bin#;
This approach however can include more bins than necessary. If there are more bins from the youngest date, than they are needed to just satisfy the target quantity, all of them will be included anyhow. So the person who picks the items for the order would have to chose from these bins. But at least the FIFO rule is kept that way and the person has to count the items anyway and cannot just blindly pick from the returned bins.
SQL Fiddle (Note, that I added bin 20 to demonstrate the above mentioned problem.)
The problem I mentioned about 1. can be circumvented if you give all the bins a number ordered by the date. Then there will be no duplicate values as with the date. You can introduce this number by using ROW_NUMBER() in a CTE. Then select from the CTE with the same logic as in 1. but applied on the row number instead of the date.
WITH cte
AS
(
SELECT ROW_NUMBER() OVER (ORDER BY b.date) row#,
b.*
FROM bin b
)
SELECT *
FROM cte c
WHERE c.row# <= (SELECT min(cc.row#)
FROM cte cc
WHERE (SELECT sum(ccc.qty)
FROM cte ccc
WHERE ccc.row# <= cc.row#) >= 50)
ORDER BY c.date,
c.bin#;
SQL Fiddle (Note, that I added bin 20 again to demonstrate, that the problem mentioned in 1. is tackled.)
Both methods however won't necessarily yield the "optimal" set of bins. For example, there might be a set of bins, with the right dates, that exactly hold the amount of items ordered but this set is only returned by chance. There might also be a set of bins with a cardinality less that the one of the returned set.