Sum days between differing date/times progromatically - sql-server

Question

I have a query that works perfectly fine, by summing the values for a day.
SELECT CAST(fldDateTime AS DATE) AS DayValue, SUM(fldValue) AS Val
FROM [dbo].[Data.tblMeterData]
GROUP BY CAST(fldDateTime AS DATE)
ORDER BY DayValue
The problem I have is that I needed to have the data greater than midnight and up until midnight of the next day. To test I can do this I tested the following code, which helped me deal with the first lot of values that do not have a full day.
DECLARE #a DATETIME
DECLARE #b DATETIME
SET #a = CAST('2016-03-21 00:01:00' AS DATETIME)
SET #b = DATEADD(SECOND,-86399 ,#a)
SELECT #a AS a, #b AS b
SELECT CAST(fldDateTime AS DATE) AS DayValue, SUM(fldValue) AS Val
FROM [dbo].[Data.tblMeterData]
WHERE fldDateTime BETWEEN #b AND #a
GROUP BY CAST(fldDateTime AS DATE)
ORDER BY DayValue
The problem with this is that I do not want to loop through dates, as I have thousands of ID's that I need to process in this way.I cannot include midnight for the first date as the first data is recorded after midnight and the final daily reading is at midnight the following day e.g.
Date > '2016-03-20 00:00:00' AND <= 2016-03-21 00:00:00
How can I do what I need to do the following:
Find the first date for an ID and move up to the first midnight of the following day and repeat this for all following days.
Sum these values so that they are greater than midnight and up to midnight of the next day.

Answer 1

My understanding of the requirements is you are looking for the sum of a value group by meter ID and date, but for each date we also want to include the next day's values. This means each value would count in the sum for its day and the previous day.
Code:
--generate test data
declare #tblMeterData table (
[ID] [int] IDENTITY(1,1) NOT NULL,
[tblMeterData_Id] [int] NOT NULL,
[fldDateTime] [datetime] NOT NULL,
[fldValue] [decimal](18, 2) NULL,
[fldBatchId] [uniqueidentifier] NOT NULL);
insert #tblMeterData (tblMeterData_Id, fldDateTime, fldValue, fldBatchId) values
(18, '2016-12-19 23:59:59', 1.0, newid()),
(18, '2016-12-20 00:00:00', 2.0, newid()),
(18, '2016-12-20 00:30:00', 3.0, newid()),
(18, '2016-12-20 01:00:00', 4.0, newid()),
(18, '2016-12-20 01:30:00', 5.0, newid()),
(18, '2016-12-21 00:00:00', 6.0, newid()),
(18, '2016-12-21 00:30:00', 7.0, newid()),
(18, '2016-12-22 00:00:00', 8.0, newid()),
(18, '2016-12-23 00:00:00', 9.0, newid()),
(19, '2016-12-20 00:00:00', 10.0, newid());
--select * from #tblMeterData order by ID;
--main query
with cte as (
--0:00:00 reports on previous day
select *, cast(dateadd(S, -1, fldDateTime) as date) group_date
from #tblMeterData
union all
--duplicate all records to also group on previous day
select *, cast(dateadd(D, -1, dateadd(S, -1, fldDateTime)) as date) group_date
from #tblMeterData
)
select tblMeterData_Id, group_date, sum(fldValue) sum_value
from cte
group by tblMeterData_Id, group_date
order by tblMeterData_Id, group_date;
Results:
tblMeterData_Id group_date sum_value
18 2016-12-18 3.00
18 2016-12-19 21.00
18 2016-12-20 33.00
18 2016-12-21 24.00
18 2016-12-22 9.00
19 2016-12-18 10.00
19 2016-12-19 10.00

Answer 2

The answer is as follows:
WITH cte AS (
SELECT *, CAST(DATEADD(S, -1, fldDateTime) AS DATE) group_date
FROM [dbo].[Data.tblMeterData]
)
SELECT tblMeterData_Id, group_date, SUM(fldValue) sum_value
FROM cte
GROUP BY tblMeterData_Id, group_date
ORDER BY tblMeterData_Id, group_date;
It works perfectly.